ROUTE WITHIN A TRANSPORTATION NETWORK

Field of the Invention

The present disclosure relates to navigation systems and road safety, and in particular, but not exclusively to improved navigation systems and methods for determining a safer route for a traffic participant to a desired destination. Aspects of the invention relate to a navigation system, to a method of determining a safest route within a transportation network, to an apparatus for determining a safest route within a transportation network, and to a vehicle comprising apparatus for determining a safest route within a transportation network.

Background to the Invention

Road safety has improved dramatically in all highly motorised countries over time, as infrastructure, vehicle safety, legislation and enforcement, driver behaviour, etc. improved. However, accidents still occur due to risks inherently associated with road traffic (e.g. road layout, etc.) as well as risk increasing factors associated with a road user (e.g. ability, fatigue, substances use, etc.).

Systematic analysis of risk factors and accident statistics has been used to identify ways of reducing the occurrence of accidents by e.g. identifying areas associated with high accident densities and singling them out for improvement, identifying drivers/behaviours and vehicle features associated with increased risk, etc. However, these approaches have been met with a limited success, partially because they provide specific remedies to single situations, rather than a general solution.

In particular, there is a need for specific knowledge of the underlying causes of crashes to be identified and used to inform and motivate road users to adopt a safe behaviour.

Navigation systems are commonly used to provide users with guidance as to the most appropriate road to be used to travel to a destination, based on user selected preferences such as time, and distance. Thus, known navigation systems are able to provide a user with the shortest route, or the quickest route between a starting location and a destination location. However, they do not take into account safety considerations in providing guidance to a user.

It is an object of the invention to improve road safety by providing navigation system capable of a determining and providing a traffic participant with the safest route between an origin and a destination location, and by providing targeted and personalised advice to the road user in order to reduce the risk associated with a trip.

Summary of Invention

An aspect of the invention provides a method of determining a safest route within a transportation network between a starting location and a destination location. The starting and destination locations may be selected by a traffic participant. The method may comprise: determining two or more potential routes, each potential route being located between the starting location and the destination location; dividing each route into one or more route segments; determining a risk coefficient associated with each one of the one or more route segments; calculating a risk factor for each one of the two or more potential routes by aggregating the risk coefficients associated with the one or more route segments comprised in each potential route; and determining the safest route on the basis of the risk factor associated with each one of the two or more potential routes. In this way, advantageously, the safest route between a starting location and a destination location may be determined. By dividing potential routes into one or more route segments enables particularly dangerous route segments to be identified and if necessary avoided. The present method may be employed by any type of traffic participant including, but not limited to motor vehicle drivers, bicycle riders, and even pedestrians.

The method may further comprise notifying the traffic participant of the safest route.

In certain embodiments, determining the safest route may comprise comparing the risk factors associated with the two or more potential routes. The route with the lowest associated risk factor may then be selected as the safest route. The use of risk factors provides a quantitative means enabling different potential routes to be compared in an objective manner.

The dividing of each route into one or more route segments may comprise identifying traffic functions along the route, and separating the route into the one or more route segments. The separating may be carried out such that any two contiguous route segments may be associated with different traffic functions. The separation of a route into component segments defined by traffic function, provides a convenient way of analysing the different potential sources of risk present in a route. For example, it is known from observation that intersections tend to be associated with a higher number of accidents when compared to a straight road segment. This is in part due to the traffic function of an intersection where two or more roads, carrying traffic moving in different directions, meet. Accordingly, there is an increase in the potential sources of an accident. A further advantage of separating a route into route segments by traffic function, is that in many countries, national traffic authorities collate accident data by the traffic function of route segments. Accordingly, this provides a convenient way of incorporating existing accident statistic data into the present method for determining a safest route.

In certain embodiments, the risk coefficient associated with a route segment may be dependent on historical accident data associated with the route segment.

In further embodiments, the risk coefficient associated with a route segment may be dependent on one or more characteristics of the route segment. The one or more characteristics may comprise any one or more of:

a) the layout of the route segment;

b) the illumination conditions of the route segment;

c) whether the route segment comprises an intersection; d) whether the route segment is located in an urban environment or in a rural environment;

e) the presence of any temporary engineering works on the route segment; f) the presence of any sharp bends in the route segment and the curvature of the bends;

g) the number of lanes present in the route segment;

h) the width of any lanes present in the route segment;

i) a gradient of the route segment; and

j) a condition of the surface of the route segment.

The above factors are all known from empirical data to contribute to the danger and therefore risk associated with a route segment.

The risk coefficient associated with the one or more route segments may be dependent on a volume of traffic associated with the one or more route segments. The greater the volume of traffic on a route segment the greater the number of potential traffic participants one can collide with.

The risk coefficient associated with the one or more route segments may be dependent on an environmental condition associated with the one or more route segments. Environmental conditions, in particular weather conditions, can have a significant impact on the safety of a route. By making risk coefficients associated with route segments dependent on an environmental condition associated with the route segment, means advantageously, that the present method can account for increases in risk associated with for example, adverse weather conditions, such as heavy rain, or even ice. Furthermore, because the environmental condition is associated with individual route segments, this means that the present method is even able to account for very local environmental phenomena which may only affect a subset of the total route segments present in an entire route. Thus, for example, the present method is able to account for local weather phenomena such as fog which may be localised at a very specific route segment, rather than being present along an entire route.

In certain embodiments, the risk coefficient associated with the one or more route segments may be dependent on a driver profile. The use of driver profiles enables the present method to account for characteristics of a driver, and thus provide a more customised route recommendation, which accounts for any shortcomings of the driver. For example, if the driver suffers from an optical condition making it more difficult for the driver to see in the dark, the present method can account for this on the basis of data comprised in the user's profile such that well illuminated route segments are prioritised when selecting the safest route.

In certain embodiments, the risk coefficient associated with each route segment comprised in each potential route is dependent on an accident prediction model. The use of accident prediction models is advantageous where there is a deficiency in accident data for a route or route segment. The use of accident prediction models enables a quantitative value of risk to be attributable to route segments, and thus facilitates the quantitative comparison of risk associated with different potential routes.

The accident prediction model may comprise calculating an estimate of the number of accidents expected to occur on each route segment as a function of the traffic each route segment is expected to experience in a time period. Similarly, the estimate of the number of accidents expected to occur on each route segment may be calculated as a function of an average daily amount of traffic each route segment is expected to experience.

In certain embodiments, the risk coefficient associated with each route segment may be calculated using an equation of the form:

μ = ctLP ^■ AADTV ^■ e∑f-i *f*i

where μ is the expected number of crashes for the route segment; L is a length of the route segment; AADT is the average annual daily traffic; x _t relate to crash modification variables which relate to characteristics of the route segment; α,β, δι and γ are coefficients to be estimated. Equations of this form are advantageous for use as they take into consideration specific characteristics of a route segment, and thus enable more accurate risk coefficients to be calculated, which are customised to the characteristics of the subject route segment.

In certain embodiments, a modified risk coefficient is calculated for each risk coefficient. The modified risk coefficient may be calculated by adapting an associated risk coefficient with a risk modification factor associated with a user profile of the traffic participant. This enables a safest route to be determined on the basis of the user profile, and advantageously provides for better user-customised results.

In certain embodiments, a performance of a traffic participant is monitored over a time period whilst the traffic participant navigates within the transportation network. A risk modification factor is generated on the basis of the monitored performance, and the generated risk modification factor is associated with the user profile of the traffic participant. In this way it is possible to automatically generate risk modification factors for specific traffic participants without requiring any input from the traffic participant. Over time, the user profile associated with the traffic participant will more accurately represent the traffic participant, and accordingly it will be possible to more accurately determine the safest route for the specific traffic participant. The use of artificial intelligence, for example, genetic algorithms, may be particularly suited for this application, since these relate to self-improving algorithms over time.

The risk modification factor may be dependent on any one or more of:

a) a number of accidents the traffic participant has previously been involved in; b) an age of the traffic participant;

c) a gender of the traffic participant;

d) the experience the traffic participant has in navigating within the

transportation network;

e) a monitored likelihood of the traffic participant exceeding a speed limit;

f) a health condition of the traffic participant; g) a level of fitness of the traffic participant; and

h) a disability of the traffic participant.

In certain embodiments, the method may comprise: receiving information associated with an update regarding a risk coefficient; determining if the risk coefficient is associated with a route segment comprised in a current route of a traffic participant; determining if the update results in an increase in the risk factor of the current route, by determining if the update relates to an increase in the risk coefficient; and providing a notification to the traffic participant if it is determined that the update results in an increase in the risk factor of the current route. This is particularly advantageous if mid-way through a journey an event, such as an accident for example, increases the risk associated with a route segment comprised in the traffic participant's selected route. By notifying the traffic participant of the increase in risk, the traffic participant is alerted to the increase in risk, and is therefore in a better position to handle the increased risk. In certain circumstances, the traffic participant may even take evasive action if necessary.

In certain embodiments, the information associated with an update regarding a risk coefficient may be received from a second traffic participant navigating within the transportation network. Enabling the sharing of information between different traffic participants within a transportation network is advantageous, in that it enables traffic participants to make more informed decisions on more up-to-date information.

Similarly, information associated with an update regarding a risk coefficient may be received from a transportation network authority. Many countries now comprise centralised transportation network authorities that maintain up-to-date overviews of the status of a transportation network, and therefore using the up-to-date information available to these authorities improves the accuracy of the present method.

Where the method of the present invention is used in motor vehicles and the transportation network relates to a road network, the transportation authority may relate to a road network traffic authority.

In certain embodiments, the method may comprise: receiving information associated with an update regarding a risk coefficient; updating the risk coefficient associated with the received information; determining if the updated risk coefficient is associated with a route segment comprised in a potential route located between a current location of the traffic participant and the destination location; calculating an updated risk factor for the potential route associated with the affected route segment;

determining if a current route is the safest route by comparing the risk factor associated with the current route, with the updated risk factor of the potential route associated with the affected route segment. This enables the present method to be more flexible and to be adaptable to compensate for any events which may impact the determination of the safest route, whilst a traffic participant is already traveling along a route. Should the event result in an alternative route now being the safest route, the traffic participant may be rerouted to the safer route. The method may comprise: determining if a change in risk factor of the potential route associated with the affected route segment is greater than a predetermined threshold, when the current route is no longer the safest route, the determining comprising comparing the updated risk factor of the potential route with a previous risk factor of the potential route associated with the affected route segment; and designating the potential route associated with the updated risk factor as the safest route, when the change in risk factor of the potential route is greater than the predetermined threshold. The use of a threshold condition ensures that any changes in the risk factor associated with potential routes, including the current route, is substantial enough to warrant notifying the traffic participant, and/or rerouting the traffic participant. This also ensures that the traffic participant is not continuously disturbed as a result of minor changes in the risk factors associated with the potential routes, which in time would dilute the effect of the present method on the traffic participant. Thus, by reducing the amount of notifications to the traffic participant, ensures that when notifications are provided to the traffic participant, the traffic participant is more likely to adhere to them, rather than dismissing them.

A further aspect of the invention provides an apparatus for determining a safest route within a transportation network between a starting location and a destination location, the starting and destination locations being selected by a traffic participant. The apparatus may comprise: an analysis module arranged to determine two or more potential routes, each potential route being located between the starting location and the destination location, and to divide each potential route into one or more route segments; a risk coefficient determining module arranged to determine a risk coefficient associated with each one of the one or more route segments; a

calculating module arranged to calculate a risk factor for each one of the two or more potential routes by aggregating the risk coefficients associated with the one or more route segments comprised in each potential route; and a route determining module arranged to determine the safest route on the basis of the risk factor associated with each one of the two or more potential routes. This aspect of the invention provides the same advantages as set out previously in relation to the previous aspect (i.e. as set out in relation to the afore-summarised method).

The route determining module may be arranged to compare the risk factors associated with the two or more potential routes, and to select the route with the lowest associated risk factor as the safest route.

The analysis module may be arranged to identify traffic functions along each potential route; and the dividing of each potential route into one or more route segments may comprise: separating each potential route into the one or more route segments, the separating being carried out such that any two contiguous route segments may be associated with different traffic functions.

The risk coefficient determining module may be arranged to determine the risk coefficient associated with a route segment in dependence on historical accident data associated with the route segment. The risk coefficient determining module may be arranged to determine the risk coefficient associated with a route segment in dependence on one or more characteristics of the route segment.

The one or more characteristics may comprise any one or more of:

a) the layout of the route segment;

b) the illumination conditions of the route segment;

c) whether the route segment comprises an intersection;

d) whether the route segment is located in an urban environment or in a rural environment;

e) the presence of any temporary engineering works on the route segment; f) the presence of any sharp bends in the route segment and the curvature of the bends;

g) the number of lanes present in the route segment;

h) the width of any lanes present in the route segment;

i) a gradient of the route segment; and

j) a condition of the surface of the route segment.

The risk coefficient determining module may be arranged to determine a risk coefficient associated with the one or more route segments in dependence on a volume of traffic associated with the one or more route segments.

The risk coefficient determining module may be arranged to determine a risk coefficient associated with the one or more route segments in dependence on an environmental condition associated with the one or more route segments.

The risk coefficient determining module may be arranged to determine a risk coefficient associated with the one or more route segments in dependence on a driver profile.

The risk coefficient determining module may be arranged to determine a risk coefficient associated with each route segment comprised in each potential route using an accident prediction model.

The accident prediction model may comprise calculating an estimate of the number of accidents expected to occur on each route segment as a function of the traffic each route segment may be expected to experience in a time period.

The estimate of the number of accidents expected to occur on each route segment may be calculated as a function of an average daily amount of traffic each route segment may be expected to experience.

The risk coefficient determining module may be arranged to calculate a risk coefficient associated with each route segment using an equation of the form:

μ = ccL? ^■ AADT · ^δ ϊ ^χ ί

where μ is the expected number of crashes for the route segment; L is a length of the route segment; AADT is the average annual daily traffic; x _t relate to crash modification variables which relate to characteristics of the route segment; α,β, δι and γ are coefficients to be estimated.

The apparatus may comprise a modified risk coefficient determining module arranged to: calculate a modified risk coefficient for each risk coefficient, by adapting the associated risk coefficient with a risk modification factor associated with a user profile of the traffic participant, in order to enable the safest route to be determined on the basis of the user profile.

The apparatus may comprise a performance monitor arranged to monitor, over a time period, a performance of the traffic participant whilst the traffic participant navigates within the transportation network; and wherein the modified risk coefficient determining module is arranged to generate the risk modification factor on the basis of the monitored performance, and to associate the risk modification factor with the user profile.

The risk modification factor may be dependent on any one or more of:

a) a number of accidents the traffic participant has previously been involved in; b) an age of the traffic participant;

c) a gender of the traffic participant;

d) the experience the traffic participant has in navigating within the

transportation network;

e) a monitored likelihood of the traffic participant exceeding a speed limit;

f) a health condition of the traffic participant;

g) a level of fitness of the traffic participant; and

h) a disability of the traffic participant.

In certain embodiments, the apparatus may comprise: a receiver arranged to receive information associated with an update regarding a risk coefficient, the receiver may be operatively coupled to the analysis module; the analysis module may be arranged to determine if the risk coefficient is associated with a route segment comprised in a current route of the traffic participant; the calculating module may be arranged to determine if the update results in an increase in the risk factor of the current route, by determining if the update relates to an increase in the risk coefficient; and a notification module that may be arranged to provide the traffic participant with a notification if the calculating module determines that the update results in an increase in the risk factor of the current route.

The receiver may be arranged to enable receipt of the information associated with the update regarding the risk coefficient from a second traffic participant navigating within the transportation network.

The receiver may be arranged to enable receipt of the information associated with the update regarding the risk coefficient from a remotely located transportation network authority.

In certain embodiments, the remotely located transportation network authority may be a road network traffic authority. The apparatus may comprise: a receiver arranged to receive information associated with an update regarding a risk coefficient, the receiver being operatively coupled to the analysis module and the risk coefficient determining module; and wherein the risk coefficient determining module may be arranged to update the risk coefficient associated with the received information; the analysis module may be arranged to determine if the risk coefficient is associated with a route segment comprised in a potential route located between a current location of the traffic participant and the destination location; the calculating module may be arranged to calculate an updated risk factor for the potential route associated with the affected route segment; and the route determining module may be arranged to determine if a current route is the safest route by comparing the risk factor associated with the current route, with the updated risk factor of the potential route associated with the affected route segment.

The calculating module may be arranged to determine if a change in risk factor of the potential route associated with the affected route segment is greater than a predetermined threshold, when the determining module determines that the current route is no longer the safest route, the calculating module may be arranged to compare the updated risk factor of the potential route with a previous risk factor of the potential route associated with the affected route segment; and wherein the route determining module may be arranged to designate the potential route associated with the updated risk factor as the safest route, when the change in risk factor of the potential route is greater than the predetermined threshold.

The apparatus may comprise a notification module operatively connected to the route determining module, and may be arranged to notify the traffic participant of the safest route as determined by the route determining module.

In certain embodiments the apparatus is a portable electronic device, such as a portable navigation system, or a mobile telephone. Alternatively, the apparatus may be comprised in a navigation system of a vehicle.

In yet a further aspect of the invention there is provided a motor vehicle comprising the afore-described apparatus, or a motor vehicle comprising a navigation system configured to carry out the afore-described method.

It is to be appreciated that aspects of the invention provide, advantageously, a system and method to determine the safest route and to provide personalised advice to a traffic participant on the basis of an analysis of the risk factors associated with a specific stretch of a route, the traffic participant, the vehicle and/or the trip.

Within the scope of this application it is expressly intended that the various aspects, embodiments, examples and alternatives set out in the preceding paragraphs, in the claims, and/or in the following description and drawings, and in particular the individual features thereof, may be taken independently or in any combination. That is, all embodiments and/or features of any embodiment may be combined in any way and/or combination, unless such features are incompatible. The applicant reserves the right to change any originally filed claim or file any new claim accordingly, including the right to amend any originally filed claim to depend from and/or incorporate any feature of any other claim although not originally claimed in that manner.

Brief Description of Drawings

Specific embodiments of the invention will be described below, by way of non-limiting example, with reference to the accompanying figures, in which:

Figure 1 is a schematic illustration of the functional modules comprised in a navigation system communicable with an external communication network, according to an embodiment of the invention;

Figure 2 is a process flow chart illustrating a method of calculating a safest route, according to an embodiment of the invention, and as carried out by the navigation system of Figure 1 ;

Figures 3a and 3b illustrate how different types of road segment may be functionally categorised according to embodiments of the invention;

Figure 4 illustrates a distribution of fatality risk per 100 million kilometres driven by road segment category;

Figure 5 illustrates a risk factor adjustment approach based on traffic volume, according to embodiments of the invention;

Figure 6 is a schematic illustration of the functional modules comprised in a navigation system configured with real time updating capability, according to embodiments of the invention;

Figure 7 is a process flow chart illustrating a method of calculating a safest route, adopted by the system of Figure 6, according to an embodiment of the invention;

Figure 8 is a process flow chart illustrating a method of calculating a safest route on the basis of a received risk factor update adopted by the system of Figure 6, in accordance with an embodiment of the invention;

Figure 9 is a schematic illustration of the functional modules comprised in a navigation system configured with vehicle-to-vehicle communication capabilities, in accordance with an embodiment of the invention;

Figure 10 is a process flow chart illustrating a method of calculating a safest route adopted by the system of Figure 9, in accordance with an embodiment of the invention;

Figure 11 is a process flow chart illustrating a method for vehicle-to-vehicle communications adopted by the system of Figure 9, in accordance with an embodiment of the invention;

Figure 12 is a process flow chart illustrating a method of updating risk coefficients adopted by the system of Figure 9, in accordance with an embodiment of the invention; Figure 13 is a graph of experimental results showing the influence of road surface friction on the accident rate in a study conducted in the UK; and

Figure 14a is a graph of experimental data showing the risk of a crash occurring as a function of age per road category type; and

Figure 14b is a graph of experimental data showing the overall risk (for males on weekend summer nights) of a crash occurring as a function of the driver age and the road type divided by the overall risk assuming no drink-driving.

Detailed Description

The present disclosure relates to a method, system and apparatus for determining a safest route that can be used by a traffic participant to select a safest route to travel to a desired destination location. Whilst the methods, system and apparatus of the present invention may be employed by any traffic participant, such as bicycle riders, motorcycles riders, automotive vehicle drivers, pedestrians, to name but a few of the potential users of the present invention, for illustrative purposes only, the ensuing description of embodiments of the invention will be described with respect to automotive vehicle drivers, and in particular with respect to automotive vehicle drivers traveling within road traffic networks. It is however to be appreciated that this is for non-limiting illustrative purposes only.

In certain embodiments, the method may comprise determining two or more potential routes from a starting location to the desired destination location. Each potential route may then be divided into its component (one or more) route segments. The component one or more route segments relate to sections of the route, which when all the route segments are aggregated effectively define the route. For example, where the route comprises a road traffic network, the component route segments may relate to sections of road comprised in the route, including intersections present along the route. These route segments will be referred to interchangeably as road segments in the ensuing description.

j

Each component road segment may be associated with a risk coefficient. The risk coefficient may in certain embodiments represent the historical accident rate associated with the road segment as a function of vehicle volume. Similarly, the number of accidents could be defined as a function of the average vehicle-kilometres driven on the road segment. In yet further embodiments of the invention the risk coefficient per route segment may be defined as the statistical likelihood of an accident occurring on a particular component road segment. In other words, the risk coefficient associated with a particular road segment may represent the statistical likelihood of an accident occurring on the subject road segment. Embodiments of the invention are not restricted to any specific definition of risk coefficient, provided that in each embodiment the adopted definition of risk coefficient is applied in a uniform manner to all potential road segments, thus enabling a quantitative comparison to be made between different routes and their associated road segments. In embodiments where the risk coefficient corresponds to the statistical likelihood of an accident occurring per unit distance (e.g. per kilometre or mile), the information may be obtained from knowledge of the total number of crashes or fatalities recorded on a specific road divided by the length of the road. Similarly, the risk coefficient may also be a function of traffic volume. That is to say, the risk coefficient may relate to the statistical likelihood of an accident occurring on a road segment as a function of the volume of traffic using the road segment. In such embodiments, the risk coefficient may for example be expressed as the statistical likelihood of an accident occurring on the subject road segment per 1000 passing vehicles, or per vehicle- kilometre, for example.

In certain embodiments, risk coefficient data may be obtained from accident statistics collated by road safety authorities, or alternatively may be obtained using statistical modelling. For example, the risk coefficient may be calculated using accident prediction modelling.

The risk coefficient associated with a specific road segment may also be a function of a variety of different characterising features associated with the subject road segment. One such characterising feature may relate to the road segment type, and specifically to its traffic function. One non-limiting way in which road segments may be categorised by traffic function is to first characterise all road segments comprised within a road network as relating either to an urban road or to a rural road. Road segments may then be further characterised according to their traffic function. In this regard, each road segment may be functionally categorised as being any one of: a through road; a distributor road; or an access road. A through road may be defined as a major road artery such as a motorway or a dual carriageway whose functional purpose is to enable vehicles to travel over larger distances at larger vehicular speeds. . A distributor road may be defined as a road whose function is to enable access to a through road. Distributor roads are often also referred to as connecting roads. Distributor roads often connect access roads to through roads. An access road may be defined as a road, which enables vehicles to access shops, and residential housing. Within a city, the majority of roads are access roads. Vehicles traveling on access roads tend to travel at lower speeds. In this regard, distributor roads may be considered as the intermediaries between through roads and access roads. Using this categorisation, a total of six different road segment types are definable: urban through roads; urban distributor roads; urban access roads; rural through roads; rural distributor roads; and rural access roads. Each different type of road segment may be associated with a different risk coefficient. In this way it is possible to functionally characterise all road segments within a road network by road type, and to associate a risk coefficient with each road segment.

In addition, intersections present in a road network may also be characterised on the basis of the types of road segments that are connected by the intersection. In this regard it is possible to define five types of intersections on the basis of the types of road segments that are connected by the intersection: through road to through road; through road to distributor road; distributor road to distributor road; distributor road to access road; and access road to access road. Different risk coefficients may be associated with different intersections on the basis of their type. Again, the risk coefficients may be related to empirical accident data available for the different types of intersection, or may be obtained using computer modelling.

The risk coefficients associated with each road segment and/or intersection may be further customised on the basis of specific characteristics associated with the subject road segment. These types of characteristics may be considered characteristics intrinsic to the road segment. For example, specific characteristics associated with the subject road segment or intersection may further impact the associated risk coefficient. Such characteristics may relate to, for example, the layout of the road segment or intersection, whether the road segment is illuminated, and the number of road segments feeding into an intersection. Further examples of the intrinsic characteristics which may affect the risk coefficient are discussed in further detail in the ensuing description. Intrinsic characteristics of road segments and intersections may be accounted for in the calculation of the associated risk coefficient by incorporating risk modification factors into the calculation of risk coefficient. In this regard, the calculation of risk coefficient for a specific type of road segment may be viewed as first involving the calculation of a base risk coefficient, which is then further customised to the particular road segment in question by the risk modification factors, which are specific to the subject road segment.

Risk modification factors may be represented in the form of coefficients or mathematical functions that increase or decrease a risk coefficient from a base model. The base model may relate to accident statistic data associated with a road segment, for example. In this way, it is possible to conceptually define the risk coefficient associated with a specific road segment or intersection as a function of the base model and the risk modification factors. This may be represented as: risk coef ficient _segment = f(risk _base ,risk modification factors) eq.1.0 where risk _base \s the risk coefficient associated with the base model, and / is a function defining the quantitative relationship between the base risk coefficient and the risk modification factors. It is to be appreciated that a plurality of different modification factors may be incorporated into the calculation of risk coefficient for a specific road segment or intersection.

Risk modification factors may also be dependent on vehicle and/or driver data. In this regard risk modification factors may be broadly categorised as either intrinsic risk modification factors (i.e. associated with specific characteristics of a road segment) or extrinsic risk modification factors (i.e. associated with factors independent of the specific road segment). Examples of extrinsic risk modification factors may relate to driver characteristics, vehicle characteristics, and even environmental characteristics such as weather. Non-limiting examples of the characteristics of a driver which may have an impact on the risk modification factor are age, gender, driving history, driving style, the physical fitness of the driver and in particular whether the driver has any health conditions such as impaired eyesight which could have an impact on the driver's driving performance. Non-limiting examples of vehicle characteristics that may have an impact on the risk modification factors are the age of the vehicle, the mass of the vehicle, the NCAP (New Car Assessment Programme) score, features of the vehicle such as the presence of Electronic Stability Control, the types of tire used, and the general maintenance condition of the vehicle including the conditions of the brake system etc.

Intrinsic risk modification factors may, for example, include whether the subject road segment is affected by any special circumstances, such as the presence of road works; the presence of lane restrictions; the general state of repair of the road segment and in particular whether the road segment has been subject to

degradation; the presence of traffic congestion as a result of accidents or road closures; the nature of the traffic, for example whether there are a very large number of trucks or other large vehicles traveling on the road segment; road surface quality; road layout; and lighting conditions.

As the person skilled in the art will appreciate, some of the above outlined risk modification factors may be related, such that risks associated with these factors may be stratified.

Once a risk coefficient has been calculated for each road segment comprised in a route, including any required risk modification factors, an aggregated risk coefficient may be obtained for each potential route. For example, this may be obtained by summing the risk coefficients associated with each road segment comprised in the subject route. By comparing the aggregated risk coefficient associated with each potential route, it is possible to select a desired route on the basis of its associated aggregated risk coefficient. For example, the route with the lowest associated aggregated risk coefficient, and thus the safest route, may be selected.

Figure 1 shows schematically the functional modules comprised in a navigation system 2 arranged to carry out the methods of the present invention, in accordance with embodiments of the present invention. The system 2 may be included as part of the native on-board electronic system of a vehicle (e.g. an on board computer and/or electronic navigation system), or may be included in a mobile computing unit, such as a mobile phone, tablet or stand-alone navigation device.

The system 2 comprises a driver interface 4, an analysis engine module 6, a communications module 8, and a GPS module 10. The analysis engine module 10 is configured to interact with one or more data sources, which data sources may comprise any one or more of vehicle data 12, driver data 14 map data 16, road data 18 and trip data 20.

The data sources may be held locally or remotely. In the illustrated embodiment of Figure 1 , the data sources comprising vehicle data 12, driver data 14, and map data 16 are illustrated as being local to the navigation system 2, whilst the road data 18 and trip data 20 are held remotely and are accessible by the analysis engine 6 over a communication network 22 (e.g. the internet) via the communications module 8. It is to be appreciated that the illustrated embodiment of Figure 1 is for non-limiting illustrative purposes only, and alternative arrangements in which one or more of the data sources are held either locally or remotely, for example in databases or data storage units, are also envisaged. It is to be appreciated that two or more of these data sources may be combined into a single database or data storage unit containing aggregated data. In some embodiments, one or more data sources may be held at a user portable device or on a server accessible via an application executable on a portable device. Similarly one or more data sources may be comprised within an onboard computer local to a vehicle, or held remotely on a server and accessible by either or both of a user portable device or an on-board vehicle computer.

The driver interface 4 comprises means for a driver to input a journey specification (e.g. via any one or more of a touch screen, keyboard, buttons, a speech recognition module, or any other data input means known in the art), and means for outputting a recommended route to a driver (e.g. via any one or more of a display, voice output, etc.) and optionally route guidance information, such as communicating potential hazards to the driver.

The driver interface 4 may be integrated in the navigation system 2 or may be a standalone portable unit that interfaces with the analysis engine 10, for example via an application on a mobile device. In such embodiments, the driver interface 4 may be configured with wireless communications capabilities, such as BlueTooth®, WiFi, etc. Journey specifications may include the driver's desired destination and any other user defined preferences, such as specific route options, a desired time and date for the journey, and/or a desired arrival or departure time and date.

The vehicle data 12 comprises information regarding the vehicle, such as for example, type of vehicle, vehicle model, vehicle mass, year of production, mileage, maintenance information, any safety features comprised in the vehicle, and/or any other information specific to the subject vehicle.

The driver data 14 comprises information that is specific to the driver, such as the driver's age, gender, driving history, or any other characteristics of the driver that may have an influence and/or be indicative of a driver's safety profile (the driver's safety profile is discussed in further detail below).

The map data 16 may comprise road layout data that may be used by the analysis engine to determine a route between locations. The road data 18 comprises information about roads, such as the type of road, number of lanes, traffic and/or accident statistics, road surface, etc.

Trip data 20 may comprise data that is specific to a trip, for example real-time or predicted traffic information, weather data, road construction work information, etc.

The analysis engine 6 may be configured to interact with any one or more of the driver interface 4, GPS module 10, communications module 8 and data sources 12, 14, 16, 18, 20 in order to receive a journey specification from a driver via the driver interface 4, determine a series of candidate routes to complete the journey via a navigation module 6b, calculate risk factors associated with the candidate routes based on data from one or more of the data sources via a risk attributing module 6c, and output one or more candidate routes with an associated risk assessment via the driver interface 4, to a driver. Operation of the navigation system 2 is further illustrated in Figure 2.

Figure 2 illustrates a method for determining a safest route, in accordance with an embodiment of the present invention. The method commences at step 200, wherein the analysis engine 6 receives a journey specification from a user via the driver interface 4. At step 202, the navigation module 6b calculates one or more candidate routes satisfying the received journey specification, using data from the map data source 16. Once the one or more candidate routes have been calculated, the risk attributing module 6c obtains data from the one or more data sources 12, 14, 18, 20 about each segment in the candidate routes and calculates 206 risk coefficients for each risk factor available, for each segment of road comprised in the candidate routes. At step 208, the analysis engine 6 determines an overall risk associated with each candidate route and selects one or more routes to display for selection. The selection may be manually selected by the driver, or the selection may be automated (e.g. the system may automatically select the route with the lowest associated risk). The analysis engine 6 then outputs 210 the one or more routes via the driver interface 4, and optionally risk information (in particular when multiple routes are output) to a driver. In particular, the input/output management module may output the route with the lowest overall risk, or a set of routes and their associated level of risk, based on which a driver may select a route to follow.

The calculation of risk coefficients will now be described in further detail.

Risk Calculation

• Road segmentation and categorisation

Risk coefficients are calculated road segments, as previously described. Road segments may comprise sections of roads, as well as intersections or junctions. Each road segment comprised in a road network may be categorised into a functional category, relating to the transport function of the subject road segment. For example, a Safe Systems approach as that developed in the Netherlands (see http://www.swov.nl/rapport R-2005-05.pdf) may be used. A road network may then be viewed as a collection of functionally categorised road segments. Alternative categorisation systems may also be used, and such alternative categorisation systems will be known to a person skilled in the art, and for this reason are not discussed in further detail.

Figures 3a and 3b illustrate an approach for categorising road segments as adopted in embodiments of the invention. Road segments may be classified in one of three different categories: through roads that facilitate traffic flow (e.g. motorways, M and A roads in Great Britain), access roads that allow access to specific destinations (e.g. streets in a town or village, minor roads), and distributor roads that provide a transition between through and access roads (e.g. collector roads, B roads in Great Britain, major roads). Each category may then further be distinguished based on whether a road segment is located in a rural or urban environment. Through roads may be broadly similar across rural and urban areas (albeit possibly with differences as to the distances between junctions), such that no distinction may be required within this category. Parameters such as access to the road by vulnerable road users (e.g. pedestrians, cyclists) may be a critical aspect of safety, and is typically a requirement for roads with a flow/through function. Alternative categorisations may be used, and such alternatives fall within the scope of the present invention. It is envisaged that different categorisation schemes may be used in different countries, for example.

Intersections and junctions may be classified according to the types of road sections that they connect. For example, intersections may relate to any one of the following types: flow-flow (grade separated intersections), flow-distributor (entry/exit to a flow road), distributor-distributor (four or three leg intersections or roundabouts), distributor-access (four or three legs intersections or roundabouts), access-access (part of a local network of streets or roads with access function only, e.g. residential area).

As the person skilled in the art would understand, different road categorisation schemes may be employed, and the scheme illustrated herein is but one example of a road categorisation scheme, and is not limiting to the scope of the present invention. The herein described methods and systems may work with alternative road categorisation schemes.

• Calculation of road segment risk coefficients

Once a road has been divided into its component road segments and features of the segments that may be associated with a risk value have been defined, a risk coefficient may be defined for each road segment.

For example, a risk coefficient may correspond to the number of crashes recorded on a road segment (possibly limited to crashes involving serious injury) divided by the distance travelled on a section of road, or divided by the number of vehicles having passed through an intersection or junction. Such data may be obtained, for example, wholly or in part from law enforcement or other government agencies, or from road users, adopting the present system and method. Traffic volumes per road segment may be measured (e.g. by road or traffic authorities using induction loop detectors, or using data from mobile devices located in vehicles) or estimated (e.g. using traffic models). Risks may be defined for specific road segments or for road segment categories. For example, the data mentioned above may be collected for given types of road, and the risk coefficient associated with each road segment will vary depending on the type of road of the segment. (In such embodiments, the feature of the road segment that is used to determine which risk coefficient applies will be the road type. For intersections, risk coefficients may also be specified according to manoeuvre type, such as e.g. turning left or right. For example, if a road segment along a route comprises an intersection, and to travel that route it is known that the driver will have to turn right at that intersection, then the risk factor associated with this manoeuvre may be higher than the risk associated with traveling through the same intersection without turning.

Figure 4 illustrates the risk of fatalities by road category. In particular, the plot in figure 4 shows the number of fatalities per 100 million kilometres driven for different road categories (through/flow: motorway/express road; distributor: general access rural/major urban; and access: limited access rural/residential street). Risks related to injuries rather than fatalities are expected to show differences according to the road category, as e.g. urban streets carry higher risks of injury than rural roads.

Therefore, compound risks taking both of these types of data into account may also be used. Example 1 below illustrates the influence of road types on risk factor calculation.

Crash prediction models may be used to compute a risk coefficient based on parameters of the road segment, when detailed data about the number of crashes recorded on the segment is not available. A crash prediction model is a mathematical model that predicts the number of crashes expected along a segment of road, as a function of parameters associated with the segment. Example 2 discussed below illustrates the use of a crash prediction model for a road section. Other crash prediction models may be used, and similar formula may be used for the expected number of crashes at intersections.

Once a risk coefficient has been obtained for a particular route segment, a variety of risk modification factors may be included in the risk coefficient calculation model. These risk modification factors may relate to the road itself (design, presence of light, etc.), the traffic on the segment, the vehicle, the driver or the trip itself (i.e. special conditions that apply to a specific trip at a specific time, such as weather, road blocks etc.). These risk modification factors that may be included in a risk coefficient calculation model will now be explained in more details.

• Road and traffic related risk modification factors.

Risks coefficients may, in some embodiments, additionally be modified based on traffic volumes for particular segments of road. For example, there is evidence that risk may decrease with increasing traffic volumes, and therefore a road type specific risk coefficient could be further improved by including data on traffic volume for a particular road segment. Figure 5 illustrates crash density as a function of annual average daily traffic (AADT). Data on the crash density as a function of traffic volume may be used to associate a different risk coefficient depending on the known or predicted traffic along a segment. Crash density (i.e. crashes per kilometre of road) may be related to traffic volume (i.e. annual average daily traffic, AADT) on different types of roads, such as urban and rural roads. The risk associated with a particular road segment may be inferred from such data by dividing the crash density by AADT. This may be used to estimate an improved risk coefficient value associated with a specific road segment, which improved risk coefficient, is now dependent on both road type and traffic volume. Similar approaches may be used to determine improved risk coefficients associated with road intersections. In some embodiments, the models used for intersections may additionally include terms related to exposure to risk.

It is to be appreciated that traffic density along certain road segments is a function of time, that is to say during certain times of the day congestion on a particular road segment will be greater than at other times, for example during rush hour. In some embodiments, the temporal dependency of traffic density may be taken into account in determining the risk associated with a specific road segment. This will be further detailed below in the "Trip related modification factors" section.

Where information about the number of accidents per hour and the hourly traffic volume is available, the hourly traffic volume may be used to calculate a risk coefficient associated with the affected road segment, instead of the average volume per day (AADT). The person skilled in the art will appreciate that the relationship between accident density per unit road length and hourly traffic volumes may be found in the existing literature (see for example Reurings, M.C.B. & Janssen, S.T.M.C. (2006), "De relatie tussen verkeersintensiteit en het aantal

verkeersongevallen voor verschillende wegtypen" which translates as "The relation between traffic volume and number of crashes for various road types; An overview of traffic models," R-2006-22. SWOV, Leidschendam).

The risk associated with specific road segments may, in some embodiments, be further expanded to include information related to road design, in addition to road type and traffic volume. For example, the below equation 2.0 may be used to quantify the expected number of crashes as a function of both the traffic volume (AADT) and road design characteristics: μ = a ^■ LP ^■ AADT ^Y■ eq.2.0 where μ is the expected number of crashes on a road segment; L is the length of the road segment in metres; AADT is the average amount of daily traffic on that road segment; x, are other characterising variables (road characteristics, such as roadway width, or number of exits - sometimes called Crash Modification Factors (CMFs) if they are part of a treatment to modify the road environment); a is a constant; β, δ ^" , and Y are coefficients to be estimated in the model. Equation 2.0 is incidentally adopted by the American Association of State Highway and Transportation Officials (the interested reader is referred to the Highway Safety Manual available at www.highwaysafetymanual.org for further information in this regard).

The road design risk modification factors may be obtained for particular designs by fitting various models to data. For example, by measuring the number of crashes in a road network, where each road comprised in the network has been categorised by road type, together with traffic volumes and road design characteristics, it is possible to attribute a risk coefficient to every segment in the road network, that takes these risk modification factors of the road segment into account.

In some embodiments, data related to road layout characteristics, traffic engineering and traffic control measures, and road construction characteristics (e.g. quality and nature of the road surface, etc.) may be incorporated into the calculation of the associated risk coefficient. It is envisaged, that in such embodiments, a data source of road segments characteristics may be available, quantifying the relationship between road characteristics and risk modification factors. This data may be generated empirically by studying the relationship between specific road attributes and accident likelihood and/or severity of accidents using Crash Prediction Models. Alternatively, this data may be obtained from the known art, for example in the Highway Safety Manual (AASHTO, 2010). Road characteristics data may be available from surveys as well. Any road characteristics that may influence a risk coefficient may be used. Examples 3a and 3b, discussed below illustrate the use of road lighting and road surface friction as risk modification factors.

Other road related modification factors that may be included in a risk calculation process may relate to the known or expected behaviour of the users of a particular road segment. Examples 4a and 4b discussed below illustrate the influence of speed and left hand turning at intersections (when driving on the right, right hand turning when driving on the left) on risk coefficients. In particular, knowledge of the speed at which users normally drive on a segment of road (whether that is determined by the speed limit or data collected e.g. from road users) may be used to calculate a risk modification factor. For example, road segments where the travel speed is lower may be associated with modification factors that result in a risk decrease over a base level, whereas roads where travel speeds are higher may result in increased risks. As will be further explained below, road related modification factors may be different depending on the road user concerned (see also Example 4b). Therefore, in some embodiments, multiple modification factors or parameters may be available and information regarding a road user may be used to determine the appropriate risk modification factor or parameter to be used in any of the predictive models mentioned above.

• Vehicle related risk modification factors

When two or more road users collide in a crash, the kinetic energy of the vehicles has a significant impact on the severity of the crash outcome. Disparities in the vehicle masses and structures between vehicles and between vehicles and pedestrians define which part of the released energy is transferred to which of the colliding partners: the smaller mass takes more of the remaining energy than the larger mass. Another determining aspect of crash severity is referred to as

"crashworthiness", i.e. the ability of a structure to protect its occupants from injury following an impact. Crashworthiness is commonly rated in laboratory tests and results are presented in star rating systems in New Car Assessment Programmes all over the world. These tests score injury prevention if a vehicle is involved in a crash (referred to as "secondary safety").

In addition, modern vehicles are frequently equipped with devices and systems aiming to prevent crashes, referred to as "primary safety". Examples of such systems include systems to prevent risky road use (e.g. alcohol lock, seat-belt lock and smartcard), systems that prevent dangerous actions during traffic participation such as systems for better vehicle control and systems that prevent violations (such as electronic stability control (ESC), intelligent speed adaptation, fatigue detection, collision avoidance systems and detection systems at intersections) and systems that reduce injury severity (e.g. pre-tensioning of seat belts, headrests optimization, e-call). The expected effectiveness of such systems may, optionally be further modified by how users will use the systems and adapt their behaviour when the systems are present (e.g. users may be more likely to adopt a riskier behaviour in the presence of such systems because they perceive the situation to be safer).

One way in which the above factors may be taken into consideration by the navigation system and method of the present invention is as follows. The distribution of traffic within a road network is often known by local traffic authorities. This information may be obtained using various different sources, which will be known to the reader skilled in the art. For example, via the use of traffic cameras, and induction loops to name but a few of the available means for monitoring and determining traffic distribution. Furthermore, the same means may also be used to obtain information regarding the types of vehicle which circulate on specific roads. For example, in this way it is possible to determine the roads which have a high volume of truck traffic. In this way, it is possible to construct a database of the entire road network, or significant portions thereof, which comprises information regarding the statistical makeup of the vehicles which commonly travel along each road within the road network. This information may then be used to determine a statistical likelihood of the types of vehicle one is likely to have an accident with if one were to have an accident on each road. In turn this information may then be used to determine a risk modification factor for each road, which accounts for this statistical likelihood. In this way, for example, roads which are heavily frequented by trucks, may be associated with a risk modification factor to account for the increased risk of injury if one were to have an accident with a truck on the affected road.

In some embodiments, some or all of the above factors (secondary

safety/crashworthiness and primary safety) may be included in the calculation of vehicle related modification factors for a vehicle travelling on a route. In such embodiments, information about the vehicle (model, crashworthiness, maintenance information, mileage etc.) may be used to calculate a risk modification factor for a vehicle on a road segment or road type. In some embodiments, vehicle related modification factors may be further differentiated based on characteristics of a user. Example 5a and 5b below demonstrate the influence of vehicle and driver related parameters on safety risks.

• Traffic participant related risk modification factors

In some embodiments, characteristics of drivers may be used to calculate driver related risk modification factors. In particular, characteristics such as age, gender and driving experience may have a large influence on the risk profile associated with a driver. The risk profile of a driver may be understood to be a measure of the statistical likelihood of the driver being involved in an accident. A breakdown of accident statistics by age, gender or experience may be used to calculate a risk modification factor.

In some embodiments, personal fitness to drive data may be used to calculate a risk modification factor. For example, a traffic participant having deficient eyesight may be more likely to have an accident, and thus may be associated with a greater risk modification factor. The risk may be particularly high for specific road segments, such as poorly illuminated roads, intersections without traffic lights, and other road segments where visually impaired traffic participants may be at a disadvantage compared to their able-sighted counterparts.

In some embodiments, a user may be able to enter some parameters that may influence a driver related risk modification factor, such as the drivers age, gender, experience, level of fatigue, any component that may influence a driver's fitness to drive (also referred to as 'task capability") such as e.g. particular medical condition that could influence a driver's ability to e.g. perceive or react to a traffic situation, any substance use (e.g. a user may be able to specify whether they have consumed alcohol or substances and optionally the specifics), tiredness, stress, use of a portable device (e.g. mobile phone) etc. In some embodiments, some or all of this data may be stored in the driver data source 14, or may be manually entered by the driver via the driver interface 4. In some embodiments, these parameters may be used by the analysis engine 6 to calculate risk modification factors for particular road segments or road types. Example 6, described below, illustrates how driver related risk modification factors affect the risk associated with different road types.

• Trip related risk modification factors

In some embodiments, characteristics of a specific trip performed at a specific time may be used to calculate a trip related risk modification factor for a road segment or road type. Such characteristics generally relate to the conditions of a trip, and may comprise weather information, specific traffic conditions that apply to a particular trip (e.g. traffic update data relating to traffic density, accidents, road blocks, etc.), road works, etc. For example, localised weather or traffic events may result in a risk increase on a particular road segment simply based on their location. Weather or traffic conditions may instead or in addition result in different risk modification factors for different segments or types of roads, even if the conditions are uniform for all the different road segments. For example, rain may make some road surfaces more slippery than others, or may have more of an impact on risk in environments where a driver is faced with greater traffic density and may have to be more vigilant to other driver behaviour (e.g. in rural vs urban environments). Some roads may similarly be more prone to adverse weather conditions, such as the formation of black ice during low temperatures. For example, roads in urban zones are more likely to be gritted and cleared of snow and/or ice than roads in rural zones.

The time of the planned trip may also influence the calculation of a trip related modification factor, as e.g. some road segments or types of roads may be associated with increased risk during the day or night time, during or outside of peak hours etc. Therefore, in some embodiments, data relating to a particular trip, such as weather, traffic data, event data (e.g. planned road works, etc.) at the time for which a trip is planned may be used to calculate trip specific risk modification factors for road segments.

As mentioned above, traffic density along certain road segments may vary as a function of time, that is to say during certain times of the day congestion on a particular road segment will be greater than at other times, for example during rush hour. Therefore, in some embodiments, the intended time of travel along a road segment may be used and incorporated into the determination of the risk coefficient associated with the road segment. In some embodiments, information regarding the day of the week, date or time of year may be comprised in the determination of the risk associated with a route. For example, in some embodiments, different predicted traffic density AADT or hourly traffic volumes may be used depending on whether a route is planned for a week day or a weekend day, for a particularly high traffic day (such as for example before a bank holiday weekend, or other public holiday ).

Real time update capability

In some embodiments, the system and method of the invention may also include dynamic (e.g. real-time) update capability, in which risk coefficients may be updated in real-time.

The functional components of a navigation system 2' configured with real-time update capability are illustrated in Figure 6. The navigation system 2' of Figure 6 shares all the functional components of the navigation system 2 of Figure 1 , and therefore like numerals albeit demarcated with an accent (2→ 2') are used to refer to like components. For the sake of conciseness and to avoid repetition, shared functional components will not be explained in detail here.

As with the navigation system 2 of Figure 1 , the navigation system 2' of Figure 6 may reside within an on-board vehicle navigation system, or may be implemented in the form of a portable electronic device. The navigation system 2' of Figure 6 may additionally comprise a driver and vehicle alert system 28, and one or more vehicle control systems 26. It is to be appreciated that whilst Figure 6 shows functionally that the one or more vehicle control systems 26 are comprised in the navigation system 2', where the navigation system relates to a stand-alone portable electronic device, the portable electronic device may be configured to interface with the vehicle control systems native to the vehicle. In addition, the navigation system 2' is configured to communicate with a central notification system 24 via a shared communication network 22'.

The central notification system 24 may relate to a data repository comprising up-to- date road network information, which information may influence the safety of a route. For example, the central notification system 24 may be arranged to share information regarding real-time traffic density, traffic incidents, road works, weather etc. with the navigation system 2'. In certain embodiments, the central notification system 24 may relate to weather or traffic update services, local authorities, etc. In certain embodiments information obtained from the central notification centre may be used to update the road data 18' and/or trip data 20'. Such real-time information may therefore be used in the calculation of risk coefficients. Therefore, in some embodiments, the data used to calculate risk coefficients and/or the data stored in the data sources 12', 14', 16', 18', and 20' may be updated as and when new data becomes available.

In some embodiments, real-time data may also be obtained by any on-board vehicle sensor and/or manually input by a user. For example, temperature sensors native to a vehicle may be used to obtain real-time weather information. On-board

temperature sensors may provide information regarding when the exterior temperature has fallen below 3°C and may be indicative of a heightened risk of black ice being present on the road. Similarly other weather sensors present on the vehicle may be used to indicate whether it is raining or snowing, for example. Data derived from on-board vehicle sensors may be incorporated into the calculation of the risk coefficients. Similarly, sensors within a vehicle passenger cabin may provide realtime information regarding a state of the driver, which can be used in the calculation of driver risk modification factors. For example, a driver facing camera may provide real-time information regarding a level of tiredness of the driver, or may provide information regarding whether the driver may be distracted, for example by a mobile telephone. This type of information may be used in the calculation of risk coefficients and/or risk modification factors.

Additionally, in some embodiments, the analysis engine 6' may also be in operative communication with the vehicle control systems 26. Information from the vehicle control systems 26 (e.g. speed provided by a cruise control system, grip of the road provided by an ABS, etc.) may be used for example to calculate vehicle related risk modification factors, and conversely information from the analysis engine 6' may be used by the vehicle control systems 26, for example to adapt the parameters of a cruise control system or automatic pilot setting.

In some embodiments, a drive and vehicle alert system 28 may be operatively connected with the analysis engine 6' and may alert the driver about external risks identified by the analysis engine 6'. This may be particularly advantageous in situations where an alternative route may not be available. In such embodiments, information about events or any features of a route that may be associated with an increased risk may be output to a driver when approaching the relevant road segment.

Figure 7 illustrates an exemplary method of calculating a route in accordance with embodiments of the invention, as may be carried out by the system of Figure 6, configured with real time update capability. The majority of the steps (200' - 210') are similar to the steps present in Figure 2, and thus the same explanations as presented previously in relation to steps 200 to 210 of Figure 2 apply. For conciseness, only the distinguishing steps, which relate to the additional real time updating capability, are now described.

After a route is output to a driver (or a route from a selected subset of routes output is chosen by a driver) the analysis engine 6' monitors the risk coefficients associated with the plurality of possible routes between the vehicle's present location and the final destination, and determines at step 212, whether a risk coefficient has been updated (i.e. as a result of a change in any of the risk modification factors). Where there is no change in any of the risk coefficients associated with any of the possible routes, the method simply returns to the monitoring and determining step 212.

If however, it is determined at step 212that there is a change in a risk coefficient, then a plurality of possible routes between the present location of the vehicle and the intended destination are calculated, at step 214. It is then determined, at step 216, whether the risk coefficient update affects any of the plurality of routes present between the current location of the vehicle and the desired destination. If it is determined that none of the plurality of routes between the present location and the destination are affected (or differentially affected), then the method returns to the monitoring and determining step 212, and the vehicle continues travelling along the selected route. If instead, it is determined that one or more of the routes located between the present location of the vehicle and the desired destination are affected, then the method returns to step 206' and the risk coefficients for each road segment present in the potential routes are calculated. The aggregated risk associated with every possible remaining route is calculated at step 208'. If it is determined that an alternative route is now the route associated with the lowest risk, then this route may be selected and output to the driver in step 210', so that the driver may choose to reroute to this new route. The decision to select a new safest route may also be conditioned on further conditions described below with reference to Figure 8. In certain embodiments, the decision to reroute the vehicle from the current route to the safer route may be automated. In other words, the decision to reroute may be taken by the navigation system 2', rather than providing it as an option for selection by the driver.

Figure 8 illustrates a further non-limiting example of how the navigation system 2' of Figure 6 may monitor and update the calculation of risk associated with potential routes in real time, in accordance with embodiments of the invention. The method of Figure 8 may be carried out whilst a vehicle is traveling within a road network on a selected route. The method begins with receipt of an update in a risk modification factor, at step 800. The associated risk coefficient is updated, at step 802, at the risk attributing module 6c'. The analysis engine 6' then verifies, at step 804, whether this change in risk modification factor affects a risk coefficient associated with at least one road segment present in any of the plurality of possible routes located between the vehicle's present location and the desired destination. If a road segment present in the plurality of possible routes is affected, then a route calculation update is triggered, at step 806. If the affected road segment is not present in any of the plurality of possible routes, then navigation system 2' continues navigating the driver along the current route, and the analysis engine 6' continues to monitor the receipt of risk factor updates, at step 800.

The risk coefficients of the affected road segments comprised in all possible routes between the current location and destination are calculated at step 806, as well as the aggregated risk coefficients for the affected possible routes. It is then determined, at step 808, whether the current route is still the route associated with the lowest aggregated risk coefficient. If the current route is no longer the route associated with the lowest aggregated risk coefficient, then the analysis engine 6' determines, at step 810, whether the change in the aggregated risk coefficient is larger than a

predetermined threshold.

If the aggregated change in risk coefficient is larger than the predetermined threshold, then the driver may be notified of this, at step 812, via the driver/vehicle alert system 28, and the new route associated with the lowest aggregated risk coefficient may be recommended to the driver via the driver interface 4'.

Alternatively, the driver may be notified directly via the driver interface 4', for example when no separate driver/vehicle alert system is present.

If it is determined, at step 808, that the current route is the route associated with the lowest aggregated risk coefficient, or if it is determined, at steps 808 and 810, that the current route is no longer the route associated with the lowest aggregated risk coefficient, but the overall change in risk is smaller than the predetermined threshold, then no alert may be provided to the driver, and the analysis engine 6' determines, at step 814, whether the received risk coefficient update relates to an increase in risk coefficient value (e.g. an increase in associated risk). It is to be appreciated that the method of the present embodiment requires two conditions to be met before a new route is recommended to the driver: 1 ) the current route must no longer be the route associated with the lowest aggregated risk coefficient; and 2) the change in aggregated risk coefficient is larger than the predetermined threshold. This predetermined threshold condition is present to ensure that the driver is not inundated with an excessive number of route change recommendations as a result of minor risk changes. Inundating the driver with too many warnings or recommendations could have the undesired effect of 'diluting' the value of the notifications received by the driver. Furthermore, it may also not be practical for the driver to be redirected if the relative benefit is insignificant.

In certain embodiments the predetermined threshold may be set by the driver, thus adjusting the individual required output levels. In certain embodiments, the analysis engine 6' may also comprise artificial intelligence, such as a genetic algorithm configured to 'learn' the driver's driving style and preferences by monitoring the driver's driving history. The predetermined threshold level may be set by the artificial intelligence. For example, by monitoring under what conditions a driver decides to select an alternative recommend route, the artificial intelligence may learn a driver's preferences and accordingly may define the threshold level accordingly. In this way, the driver would only be provided with recommended route alterations when they are likely to be acceptable to the driver, on the basis of observed precedent.

If it is determined, at step 814, that the risk coefficient update relates to an increase in coefficient value, the analysis engine 6' may also verify, at step 816, whether the driver's current location lies within a predetermined distance from the one or more road segments associated with the received increase in risk coefficient value. If the driver's current location is within the predetermined distance of the road segment affected by the change in risk coefficient, then the driver is notified of the change, at step 818. In some embodiments, step 814 may additionally comprise determining whether the risk coefficient update relates to an increase, which is larger than a predetermined threshold. This additional predetermined threshold condition is to ensure that the driver does not receive an excessive amount of notifications.

If the segment associated with increased risk is not located within a predetermined distance from the driver's current position, the notification is 'stored', at step 820. The method is then repeated by returning to the receiving step 800. This ensures that the notification is released once the vehicle is within the predetermined distance of the road segment affected by the change in risk coefficient.

The predetermined distance may be set by the driver, to adjust the "risk sensitivity" of the method in accordance with individual preferences. In certain embodiments, the predetermined distance value may be defined in terms of a distance from the driver's current location. Alternatively, the predetermined distance may be determined in terms of an estimated time between the driver's current location and the one or more road segments associated with the risk coefficient increase. For example, a driver may set the predetermined distance to be 5 miles. If it is determined, at step 816, that the one or more road segments associated with the received risk coefficient increase are located within an estimated 5 miles from the driver's current location, then the driver is notified, at step 818 via the driver/vehicle alert system 28. The notification may only be provided once the vehicle is within 5 miles of the one or more road segments associated with the received risk coefficient increase.

If the updated received risk coefficient is determined, at step 814, to relate to a decrease in risk coefficient value, then no driver alert is generated and the analysis engine 6' continues to monitor changes in risk coefficient values, and the method is commenced from step 800 upon receipt of a risk factor update. This iterative method enables the navigation system 2' to determine if a vehicle should be rerouted from its current route, on the basis of receipt of up-to-date risk information.

Optionally, in certain embodiments, if the magnitude of the received risk increase is very large (e.g. it relates to a steep increase in risk coefficient value, larger than a predetermined safety threshold), then the vehicle control system 26 may be activated. This may relate to activating the vehicle's speed limiter to reduce the maximum speed of the vehicle, for example.

As the person skilled in the art would understand, updates regarding changes to risk factors may be received in real-time, continuously, or at periodic intervals. For example, the process of Figure 8 may be repeated at regular intervals, defined in terms of time or distance, or may be triggered for example every time a road segment has been travelled and a vehicle is about to enter a new road segment. Similarly, the process may be triggered in the vicinity of intersections, since intersections provide an opportunity for a vehicle to be rerouted.

Vehicle-to-vehicle and infrastructure-to-vehicle communication capabilities

In some embodiments, the system may incorporate one or more of vehicle-to-vehicle communication and infrastructure-to-vehicle communication capabilities. In such embodiments, the system may be configured to receive updates and amendments of risk coefficients in real-time, or at discrete temporal intervals, using any one or more of the available communication capabilities of the navigation system.

In infrastructure-to-vehicle embodiments, information associated with risk

coefficients, for example information used to calculate changes in risk coefficients, may be received from road network infrastructure. For example, this may be achieved using vehicular communication systems, in which vehicles communicate with roadside units and/or roadside stations. Roadside units may be standalone units, located along road infrastructure, or they may be incorporated into existing network infrastructure such as traffic lights, road signs, toll booths, etc. The precise details of the vehicular communication system are beyond the scope of the present invention, nonetheless the interested reader is referred to the following website address for further information in this regard

http://en.wikipedia.org/wiki/Vehicular communication systems.

In such embodiments, traffic vehicles and roadside units and/or roadside stations may be configured with a shared communications means, such as a radio communication device configured to enable communication with the system of the invention. For example, the vehicle and the roadside units and/or roadside stations may be configured in accordance with the IEEE 802.11p standard

http://en.wikipedia.Org/wiki/IEEE_802.11 p. As the vehicle approaches the relevant roadside unit and/or roadside stations the vehicle may be provided with updated information regarding a specific risk modification factor, via the shared

communications means. An advantage of using infrastructure-to-vehicle

communication is that localised risk modification factor information may be communicated to potentially affected vehicles within the affected geographical area. In this way, only those vehicles that might be affected by the traffic information are notified.

Using infrastructure-to-vehicle communications, the methods and system of the present invention may be incorporated into an Intelligent Transportation System (ITS). Such ITS are currently being developed in many countries, and will become more prominent in the future. The present invention is well suited for integration into such systems.

Alternatively, or in addition, risk modification factor information affecting a route may be communicated directly between vehicles using vehicle-to-vehicle ("V2V") communications. In such embodiments, participant vehicles will be referred to as 'subscribed traffic participants' or 'subscribed vehicles'. The use of vehicle-to-vehicle (V2V) communications provide a further layer of security by enabling proximally located subscribed vehicles to communicate and share information associated with risks related to nearby road segments to be shared in environments where infrastructure-to-vehicle communications means are unavailable. In addition, V2V communications provide a means for vehicles in the vicinity to share route information with each other. This may subsequently be used for traffic congestion management purposes. For example, if a subject vehicle observes that a significant portion of proximally located vehicles intend to take the same route as the subject vehicle, this information could be indicative that the selected route will experience increased congestion, which in turn could be indicative of a change in risk associated with the selected route. On the basis of this received information, an alternative route associated with less traffic congestion, and therefore lower risk may be determined. In such embodiments, information acquired by a vehicle from other vehicles may be used to update the risk coefficients for a specific route, but also to update the data sources 18' and/or 20' for access to the information by other vehicles. Figure 9 shows an exemplary embodiment of a system of the invention comprising a vehicle-to-vehicle capability. The system comprises the features of the system of Figure 6, to which are added a vehicle to vehicle communications module 30 and an infrastructure communication centre 32. The vehicle-to-vehicle communications module 30 is operatively connected to the communications module 8" and is arranged to receive communication data regarding the current location of other vehicles located in the vicinity. Additionally, the vehicle-to-vehicle communications module 30 may enable notifications regarding risk modification factors affecting road segments to be exchanged between subscribed vehicles.

In order to accommodate this additional capability, the analysis engine 6" of Figure 9 comprises the additional functionality of identifying potentially dangerous events and notifying the communications module 8", which in turn notifies the vehicle-to-vehicle communications module 30, which transmits the notifications to all subscribed vehicles within a specified radius. For example, if a first subscribed vehicle suffers a breakdown and is partly blocking a road segment, an urgent notification may be sent to a second subscribed vehicle and third subscribed vehicle located in the vicinity of the first vehicle, so that the drivers of the second and third vehicles are aware of the potential increased risk associated with the broken down first vehicle partly obstructing the road. This feature is described in more detail with reference to Figure 10.

The central safety information centre 32 is operatively connected with the

communications module 8", for example via the communication network 22'. The infrastructure communication centre 32 receives updates regarding the personalised advice provided to the subscribed vehicle, updates concerning the current location of the subscribed vehicle and information from the road authorities. As explained above, this information may be used, e.g. to update road or tip data and/or associated risk modification factors.

Figure 10 illustrates a general route calculation process according to an embodiment of the invention where information is shared between subscribed participants. The majority of the steps, namely steps 200" to 212" have been described before with reference to Figure 7. Following the risk coefficient updating step 212", the current location and the currently proposed route of the vehicle are updated in step 900 in the central safety information centre 32 via the communications module 8" and communication network 22". This allows the central safety information centre 32 to have a real time image of where all the subscribed vehicles are located and what their planned future location is.

In some embodiments, the methods and systems of the present invention may take into account the uncertainty associated with the data concerning the (planned) future locations of the subscribed vehicles and compensates for that uncertainty by, for example, applying a discount coefficient to the (planned) future location data. Unlike the data regarding the current (real time) location of a vehicle, which is 100% accurate, the future (planned) location of the vehicle is merely an estimate relying on a prediction and is thus subject to a much wider error margin. This is because the future planned location of a vehicle is predicted on the basis of the journey specification that has been proposed to the subscribed vehicle and which the subscribed vehicle is thought to follow at a specific time. However, this prediction may not be accurate for a number of reasons: the subscribed vehicle may not follow the proposed route, there might be a re-routing instruction later during the journey, or the route might be delayed, leading to the vehicle being at the predicted locations later or sooner than the predicted time. In order to compensate for this uncertainty and to reflect the decaying level of confidence in the predicted future location data, an accuracy coefficient function may be applied to all subscribed vehicle location data stored in the central safety information centre 32.

By way of example, during the course of a journey which is planned to be three hours long, the location of the subscribed vehicle may technically be predicted for any point in time between the beginning of the journey (tO) and the end of the journey (to + 3). However, taking into account the decaying confidence in the reliability of the location data, the accuracy coefficient function is applied to the location of the vehicle. The confidence level is inversely proportional to how far in advance the location data is being projected; e.g. the predicted location of the vehicle five minutes into the future is much more likely to be accurate than the predicted location of the subscribed vehicle two hours into the future, and is thus associated with a higher confidence value. This accuracy coefficient may be in the form of a decaying exponential function, yielding its highest value at t=0 (present). The use of this coefficient will become clearer upon describing an enhanced risk modification factor update feature, with reference to Figure 12.

Alternatively, in some embodiments of the invention in order to compensate for this uncertainty and to reflect the decaying level of confidence in the predicted future location data, it is only the future predicted location of the subscribed vehicle within a predetermined 'cut-off time interval which is updated on the central safety information centre 32. In some embodiments, the cut-off time interval may be set by the driver on an individual basis. By way of example, driver A may choose to use traffic data only up to 30 minutes into the future, whilst driver B may choose to use traffic data up to 2 hours in to the future.

In some embodiments, the cut-off time feature may be used in conjunction with the accuracy coefficient function or without it. These three alternatives effectively result in three different accuracy coefficient profiles.

Figure 11 is a flow chart illustrating a general overview of the vehicle-to-vehicle communications feature as used in embodiments of the invention. The analysis engine 6" monitors internal risk modification factors (i.e. risk modification factors associated with a driver or vehicle, such as the condition of the tyres of the car, the health and level of impairment of the driver, etc.), in order to determine whether there is a 'sharp' increase in an internal risk modification factor. In this instance, sharpness is defined as a change larger than a predetermined threshold value, which occurs in a time interval smaller than a predetermined time interval threshold.

If a 'sharp' increase is detected, in step 100, in an internal risk modification factor, the communications module 8" updates the central safety information centre 32. It is then determined, in step 104, using information provided by t the central safety information centre 32, whether any subscribed vehicles are located within a predetermined distance from the current location of the vehicle. If so, then an urgent notification is sent, in step 106, to the vehicles that are found to be within that range ('neighbouring vehicles') via the vehicle-to-vehicle communications module 30. This urgent notification will be picked up by the neighbouring vehicles' vehicle-to-vehicle communications modules and will be output to the drivers. If the vehicle to vehicle communications module 30 receives, in step 108, an urgent notification form another vehicle, the driver is instantly alerted, in step 110, via the driver/vehicle alert system 28'. The purpose of this feature is to inform drivers of sudden risks arising within a short time window so that they can be alert and better prepared for the increased upcoming risk.

It should be noted that even in the absence of the vehicle-to-vehicle communications system, the driver would eventually receive a notification using the notification feature described previously with reference to Figure 6. However, it is to be appreciated that the vehicle-to-vehicle communication method is faster because fewer analysis steps are required, and as time is of the essence in such situations, the present embodiment provides an additional layer of safety.

The additional location feature described with reference to Figure 10 allows for the enhanced risk modification factor update feature illustrated in the flow diagram of Figure 12. As described with reference to Figure 10, the central safety information centre 32 continuously receives the current and planned future locations of all the subscribed vehicles. Using this image of the current and planned future location of the subscribed vehicles, as well as general traffic data, the central safety information centre may continuously monitor 120 the number of vehicles populating every given road segment. During the monitoring of the number of vehicles populating each road segment, it is determined, in step 122, whether a road segment is currently or is going to be, in the future, populated by a number of vehicles larger than a

predetermined threshold n.

The predetermined threshold n is a parameter which reflects the maximum traffic density for which a road segment is safe and efficient to drive through. The predetermined threshold n may be calculated centrally, at the central safety information centre 32, and may be different for every road segment, depending on parameters such as those described above in relation to the risk coefficient calculation process. The calculation of the current number of vehicles on each road segment may be based on general traffic data and the current location of the subscribed vehicles. The calculation of the future number of vehicles on each road segment may be based on the personalised journey instructions provided to the subscribed vehicles. As described previously with reference to Figure 10, an accuracy coefficient may be applied to the data regarding the locations of the subscribed vehicles. This accuracy coefficient may be a function of time in the form of a decaying exponential, such as F(t)=e-kt. It is clear that the maximum value of the accuracy coefficient function is at the present time, when t=0. Furthermore, it is also clear that even if the same number of cars m is predicted to be located at a road segment at t1 and t2, t1<t2, the traffic density will be estimated to be larger at t1 than t2.

If it is determined, in step 122, that a road segment is going to be populated by a number of vehicles larger than its corresponding threshold n, then the risk coefficient associated with that road segment is increased, in step 124, by the central safety information centre 32 and the risk coefficient update 126 is sent to all subscribed vehicles in step 128. If the number of vehicles is not determined to exceed the predetermined threshold n, then no action is taken and the system continues to monitor the location data. It should be noted that the threshold n is customised for each road segment based on its function, capacity and other factors as described above.

In some embodiments, the method of the invention may further comprise an assignment of 'vehicle identification codes' to the traffic data related to each vehicle. This may be used in order to ensure that the multiple streams of location and traffic data are integrated correctly and no vehicles are accounted for multiple times.

For example, imagining that a local department of transportation sends the information that in ten minutes, road segment A will be populated by seven vehicles. The central server of the embodiment also sends information that three subscribed vehicles are being routed through routes that result in road segment A being populated by three subscribed vehicles in ten minutes. Depending on the overlap between the two sources of data our prediction for the traffic volume in road segment A may vary between seven vehicles (100% overlap) and ten vehicles (0% overlap), which is a significant variation. Knowing which three cars are the subscribed vehicles that the central safety server suggests will be on road segment and A and which are the ten cars that the general traffic data suggest will be on road segment A means that the overlap can be easily calculated.

Optionally, the present embodiment may also periodically send the traffic data harvested from the subscribed vehicles to road administration authorities for that data to be used to enhance traffic management.

Example 1. This example illustrates the influence of road type on risk coefficients. In particular, data collected in three countries (Netherlands, Norway and the United Kingdom) illustrates the impact of road type on safety risk. As explained above, road segments may be distinguished based on whether they are road sections and intersections. In order to do this, one must decide which intersections are included in a road section (commonly minor intersections, e.g. three legs with access roads) and which are considered separately (commonly major intersections).

Table 1 below shows injury risks (injury accidents per million motor vehicle- kilometres) for road sections in the Netherlands in 1994 (courtesy of Poppe, 1996 , "Risico's onderscheiden naar wegtype," Leidschendam: SWOV Institute for Road Safety Research).

Table 1

Table 2 shows accident rates on national highways in Norway in 2005 (courtesy of Elvik et al., 2009, "The Handbook of Road Safety Measures," Emerald, Bingley).

Motorways 0.07 1

Substandard motorways 0.11 1.5

Rural roads 0.14 2

Urban streets 0.37 5

Table 2

Table 3 shows relative accident risks by road type for the UK in 2007 (courtesy of Bayliss, 2009, "Accident Trends by Road Type," London: Royal Automobile Club Foundation).

Motorways 1 1 1

Rural 'A' roads (major) 4.7 4.3 2.9

Rural minor 5.8 7.7 5

Urban 'A' streets (major) 3.8 9.3 7.4

Urban minor 2.6 8.2 6.8 Table 3

Similar data can be obtained for intersections. Table 4 below shows data from the Netherlands reflecting injury accidents per passing motor vehicle (courtesy of Poppe, 1996, "Risico's onderseheiden naar wegtype" (Differentiating traffic risks according to type of road), Leidschendam: SWOV Institute for Road Safety Research).

Rural all, signalized 0.22 3.9

Urban, three legs, no 0.092 1.6

lights

Urban, four legs, no lights 0.077 1.4

Urban, roundabout 0.056 1

Urban, three legs, lights 0.132 2.4

Urban, four legs, lights 0.147 2.6

Table 4. Injury accidents per passing motor vehicle (courtesy of Poppe, 1996, "Risico's onderseheiden naar wegtype" (Differentiating traffic risks according to type of road), Leidschendam: SWOV Institute for Road Safety Research).

This data mainly focuses on distributor-distributor intersections. As can be seen from the data above, it is of relevance to distinguish distributor-distributor intersections inside and outside built-up areas, because of the complete different road lay-out and traffic characteristics. The data above also illustrates that different intersection types have different risk levels.

When more detailed information is available, it may be used to obtain more specific estimates of risks associated with intersections. Table 5 below shows Dutch data for major rural intersections. To arrive at risk figures we have to take into account the number of vehicles that enter the different intersection types.

A similar approach can be taken if we compare four legged priority intersections with signalized intersections. The density of injury accidents for priority intersection is lower than for signalized intersection. However, this could be different if we multiply these density figures with data on traffic volumes, assuming that these volumes are higher for signalized intersections than for priority intersections. Signalized; four legs 6.5 1.5

Give way; four legs 4.2 1

Table 6. Density of injury crashes (numbers per year) on rural distributor intersections in the Netherlands comparing signalized intersections with give way intersections (courtesy of Dijkstra, 2014)

The data presented in this example illustrates that different risk coefficients may be obtained for different types of roads segments (road sections or intersections).

Depending on the data and knowledge available about segments ina road network, the quality of an estimated risk coefficient may vary, and being a specific as possible may improve the quality of recommendations to find the safest route between A and B in a network, as can be seen from the data.

Example 2. The relationship between the crash density associated with a road type and AADT is a function of the design characteristics of the road. This example illustrates the use of a base crash prediction model for a road segment, and in particular the use of equation 2.0 in crash prediction for a road segment. The nominal conditions are:

- Lane width = 3.6 m

- Shoulder width = 1.8 m

- Roadside hazard rating = 3

- Driveway density = 3 driveways per km

- Horizontal and vertical curvature = none

- Grade = level (0 percent)

The base model for the road segments is: μ = 365 1Q- ⁶ AADT L e -0.4865 where μ is the expected number of road crashes on a road segment, AADT is the annual average daily traffic and L is the length of a segment of a road.

This example illustrates how a crash prediction model may be used to calculate an expected risk associated with a road segment.

Example 3a. This example illustrates the effect of road safety interventions/attributes on risks associated with a road segment. Road safety interventions may comprise road design, equipment of roads, maintenance of roads, traffic control measures, etc. This example focuses on the effect of road lighting. The objective of road lighting is to reduce the accident rate in the dark by making it easier to see the road, other drivers and the immediate surroundings of the road. In the Handbook of Road Safety Measures (Elvik, 2009), the authors carried out a so-called meta analyses, in which a weighted (according to sample size, quality of a study, etc.) average from a collection of individual studies is used to obtain a more robust estimate of the effect size of a certain intervention. Results coming from a meta-analysis combine data from different settings and hence may be considered as a good starting point for intervention effects. If more detailed or focussed studies are available, they may be used to obtain reliable estimates of effects of interventions. The table below

illustrates the safety effects of public or road lighting, that is to say the number of injury accidents that are prevented if an unlit road will have road lighting. testes

Safety effects (%) - 4 -14 -22 -29 -40

Table 7. Effects on injury accidents of lighting of previously unlit roads (Elvik et al., 2009).

This data demonstrates how the presence or absence of road lighting can be used to calculate road related risk modification coefficients.

Example 3b. This example illustrates the effect of another road parameter on the risk coefficient associated with a road segment, in particular road friction. Road

surface friction influences a vehicle's ability to steer and brake. Road surface friction is quantified by a friction coefficient between 0 and 1. Typical values for dry bare asphalt are between 0.7 and 0.9, between 0.4 and 0.7 for wet bare asphalt, and between 0.1 and 0.4 for snow or ice-covered roads. These values may of course vary, and are just given as indications. Drivers have a tendency to reduce their speed on wet roads, but not sufficiently to offset the increase in risk due to lowered friction.

Therefore, the influence of road surface friction on road crashes is typically higher on wet roads than on dry roads.

Figure 13 shows the influence of road surface friction on the accident rate in a study conducted in the UK (Parry, A.R., Viner, H.E., 2005. Accidents and the skidding resistance standard for strategic roads in England. TRL). The authors developed crash prediction models for three categories of roads:

AB: motorway and dual carriageway non-event

C: single carriageway non-event

Event: (approaches to) junctions, roundabouts, pedestrians, gradients (>5%), and bends (with a radius under 500 m).

The results on Figure 13 show that road surface friction has an impact on the crash rate, and that this impact differs depending on the road type. In particular, the data shows that the likelihood of accident decreases as skid resistance increases, and that the impact is greater for event sites than for motorway sections (i.e. the power of the curve is -2.5 versus -0.9).

This example demonstrates that this and other similar data can be used to calculate risk coefficients for different road segments or road types, based on a friction related factor.

Example 4a. This example demonstrates the influence of speed choice on risk

coefficients. High driving speeds are known to lead to a higher crash rate, and

greater likelihood of a more severe outcome. This relationship is reflected in risk levels for road segments. If a specific road segment has a higher speed level than similar segments (for example with the same speed limit and the same road layout), then risks of these high-speeds segments will be higher. If a traffic participant has to 'follow the flow', the risk level of that road segment should be modified and increased. If travel speeds are lower, modification should result in a lower risk level. The relationship between speed and risk can be quantified, for example following the 'power law of Nilsson': if speed changes, the associated risk changes with a power function and this power is higher for a more severe outcome of an accident. More recently, it was shown that the power may be different for different road types. The data below illustrates this for serious injury accidents.

Accidents with serious 2.6 1.5 injury

Table 8. The exponents of the power functions for the relationship between speed and accidents (Elvik, 2009).

Example 4b. This example illustrates the influence of another road feature on risk associated with a road segment, i.e. left hand turns at intersections (in right hand traffic). We know from the literature that left hand turns (with right hand traffic) at intersections are relatively risky, presumably because a driver has to be aware of traffic coming from different directions at the same time, assess the speed and find a safe gap while being aware of vehicles coming from left and right, and coming from the back. Older drivers and novice drivers are known to find this manoeuvre complicated, and run a relatively high risk. The present inventors have analysed Dutch accident data to better understand the risks related to this manoeuvre and distinguished age group and road type.

Netherlands in 2000-2008 (SWOV, 2016 - httDs://www.swov.nl/ibmcoanos/cai- bin/coanos.cai?)

This data shows that the presence of left hand turns in right hand traffic increases the safety risk, and that this increase is dependent on both the type of road and the age of the driver.

Example 5a. This example demonstrates the influence of vehicle type on a risk coefficient. The data in table 10 below demonstrates that different risk are associated with different types of vehicles, and that this difference may be further compounded with differences associated with the age and gender of a driver.

Similarly, differences between types of vehicles are expected to be compounded by features of the road segment or road type (e.g. because visibility, road adherence, etc. may vary between vehicles and such factors may have different effects on different road segments or types of roads), and such effects can be studied in a similar way.

Age Gender PC SUV PU MVAN MC Overall

Young Male 22.82 14.25 1655 12.46 97.66 21.79

Mid 6.27 2.78 5.13 1.59 2339 6.00

Old 4.92 2.03 4.45 1.61 260.06 4.74

Young Female 19.39 21.18 27.04 13.51 NA 20.24

Mid 6.12 2.98 5.21 1.88 4838 5.59

Old 5.61 3.51 4.60 2.66 NA 5.52

Overall 6.81 3.22 5.62 1.89 28.62 6.40

Table 10. Crash rates of US drivers (per 106 miles driven, 1995) (courtesy ofKweon,

Y.J., Kockelman, K.M. (2003), "Overall injury risk to different drivers: combining exposure, frequency, and severity models," Accident Analysis and Prevention, 35(4),

441-450).

Example 5b. This example illustrates the effect of a vehicle related parameter on the risk coefficients for different types of roads. In particular, this example studies the relationship between single-vehicle crashes and the availability of Electronic Stability Control (ESC).

ESC most significantly affects single-vehicle crashes. On the one hand, data from Lyckegaard et al. (Lyckegaard, A., Hels, T., Bernhoft, I. M. (2015). Effectiveness of electronic stability control on single-vehicle accidents. Traffic injury prevention, 16(4), 380-386.) suggests that the presence of ESC results in a 31% reduction of single- vehicle crashes. On the other hand, the data in Table 11 shows that substantial differences exist in the share of serious injuries due to single-vehicle crashes between road types. Therefore, multiplying these shares by the 31% yields an estimated effect of ESC per road type. Additionally, it is possible to further distinguish this effect between types of vehicle. Farmer (Farmer, C. . (2006). Effects of electronic stability control: an update. Traffic injury prevention, 7(4), 319-324.) reported the effect of ESC was higher for SUVs (49%) than for passenger cars (33%). With these figures, the right column in Table 11 can be split according to an estimated effectiveness of ESC for SUVs and passenger cars. 30 21 6

50 32 10

60 54 17

70 24 8

80 42 13

Motorway 32 10

* Share of us injuries in single-vehicle crashes multiplied by an average effectiveness of ESC in single-vehicle crashes of 31%

Table 11. Share of severe injuries in single crashes in the Netherlands in 2000-2008 per road type (SWOV, 2016).

Example 6. This example demonstrates the influence of driver related risk factors on risk coefficients associated with different types of roads. In this example, the risk of crash (crashes involved per 100 million kilometre) is estimated for six road types (major urban and minor urban, three rural types and motorways) and different age groups. The results of this study are shown in Figure 14a, which shows model estimates for female drivers during the day on weekdays during autumn/spring, by age and road type (Keall, .D., Frith, W.J. (2004). Older driver crash rates in relation to type and quantity of travel. Traffic Injury Prevention, 5(1 ), 26-36). Generally, the data shows a U-shaped curve: the younger and the older the higher the risk. One road type (motorways) seems to be the exception. Risks of older and younger drivers seem to be less elevated on motorways than on other road types. Although the elderly prefer to avoid driving on motorways (it is reported that they are somewhat afraid of high speed environments), the data shows that it is safer for them.

Additionally, the data suggests that there is a strong difference between risks associated with male and female drivers, in particular for the youngest age groups up to 30 years old (at least a factor of 2 for the very young) and this has to be included as well. Such knowledge may be used to calculate a personalised risk coefficient.

The same dataset was used to examine the relationship between drink driving and risk, more specifically the contribution of alcohol to night time risks for different age groups and for different road types. This relation is depicted for males on weekend summer nights in Figure 14b, which shows the ratio of risk for males driving on weekend summer nights by driver age and road type to sober males; and the same quantity for males on minor urban roads in daytime (dashes) (Keall, M.D., Frith, W.J., Patterson, T.L. (2005). The contribution of alcohol to night time crash risk and other risks of night driving, "Accident Analysis and Prevention," 37(5), 816-824). As can been seen, the younger age groups dominate the higher risks related to drinking and driving. Furthermore, high volume roads (divided state highways and motorways) experience hardly any effect of drinking and driving problems. Example 7. This example demonstrates qualitatively how embodiments of the invention may be used to provide personalised navigation advice. First, the driver's journey specification details are received. This specification may be presented just before a trip starts, or for example one day prior to the journey. The journey specification details include the following:

The starting point of the journey is Vauxhall, London, UK, and the destination is Hillsborough Stadium, Sheffield, UK.

The route must also pass via Luton airport; the desired arrival time at

Hillsborough Stadium is 10pm on 10 November 2014; and the departure time is not prior to 4pm on the same day.

A plurality of possible routes - routes A, B, C - that satisfy the journey specification criteria are calculated by the analysis engine 6. Each possible route is divided into its component road segments and the (stored) risk modification factors are retrieved from the road data source 18 and presented to the analysis engine 6. Subsequently, a risk coefficient is determined, taking into account each risk modification factor identified by the input devices. Some of these risk factors may comprise one or more of the following:

According to information provided by the communications module 8 (from e.g. the central notification centre 24 or the trip data source 22), rain is predicted in the Sheffield area on the evening of 10 November, resulting in an increased risk factor for the affected road segments. In addition to this, there are predicted engineering works along two miles on the M1 in the vicinity of East Midlands Airport. The predicted increased traffic because of the engineering works results in an increased risk coefficient applied to all affected road segments.

The driver, according to information provided by the driver data source 14, is a 70 year old male with mild vision problems (this is an example of a global internal risk factor). As a result, and because the journey is carried out in the evening hours, when the roads tend to be less well illuminated, it is expected that the aggregated risk coefficient associated with the entire journey will increase. For example, the risk coefficient associated with each road segment will account for the driver's visual impairment, and the risk coefficient value associated with, in particular, poorly lit road segments will be correspondingly greater than the risk coefficient value associated with a well-lit road segment. For example, a dark country lane will be associated with a correspondingly larger risk coefficient value than a well-lit motorway road segment.

The vehicle, according to information provided by the vehicle data source unit 12 last had its tyres replaced 50,000 kilometres ago, and as a result the tread depth has worn down to 2.5mm. Accordingly, an increased risk factor applies to wet tarmac, and an increased risk coefficient is allocated to all road segments affected by the wet weather on the evening of 10 November.

It is to be appreciated that a variable such as weather, and specifically rain in the above illustrated example, may, in certain circumstances, affect a plurality of road segments associated with different potential routes in the same manner. This effectively results in a uniform increase in the associated risk for all affected road segments. When this is the case, the system may recommend that the driver maintain their current, existing route. However, if the different potential routes are not affected in a uniform manner, for example when the current route is affected by a very localised rain shower, or other weather event (e.g. fog), or is more affected because of lower road surface friction, then, the system may recommend rerouting to a route that is now associated with a lower aggregated risk coefficient compared to the current route.

On the basis of the risk coefficients determined by the risk attributing module 6c, the risk coefficient associated with each road segment is calculated. Subsequently, the overall aggregated risk coefficient associated with each of the routes A, B and C is calculated. The route associated with the lowest aggregated risk coefficient, - route C - is selected, and the result is output to the driver. On this basis, the driver now commences their journey.

During the journey, whilst the driver drives along route C, the analysis engine 6 continuously monitors all input devices, in order to determine and monitor changes in risk coefficients associated with the road segments of any of the plurality of calculated possible routes between the vehicle's current position and the desired destination. Shortly after the beginning the journey, a notification is received, from the communications network 22, that there has been an incident which resulted in lubricant spillage on the road surface along a segment of the M1 ('Disturbance 1'), fifty miles south of Sheffield. Therefore, the risk attributing module 6c updates and amends the relevant risk coefficients.

Following the risk coefficient update, it is determined whether the update affects any of the road segments comprised within any of the possible routes that lie between the vehicle's present location and the destination. It is determined that the affected road segment is situated along route C - e.g. along the current route. Accordingly, the aggregated risk coefficient associated with every possible route present between the vehicle's current location and the destination is recalculated. It is then

determined whether route C remains the safest route. Because route C remains the safest route despite the accident, the analysis engine 6 does not propose redirecting the driver via a different possible route.

Subsequently, it is determined whether the received risk coefficient update relates to an increase in risk coefficient value. Due to the spilled lubricant on the road, the risk coefficient update is determined to be an increase.

The proximity of the affected road segment is then determined. This corresponds to the aforementioned step of determining if the increased risk coefficient is associated with a road segment, which lies within the predetermined threshold. It is determined that the affected road segment is situated more than two hours' drive from the vehicle's present location, which is outside the predetermined threshold, such that no notification is provided to the driver. Later during the journey, a notification regarding weather conditions is received from the communications network 22', indicating a change in the originally predicted rain pattern for the journey hours ('Disturbance 2'). The risk attributing module 6c updates and amends the relevant risk coefficients. Following this, it is determined whether the change affects any of the road segments along any of the routes of the plurality of possible routes between the present location of the vehicle and the destination.

The new rain pattern predicts clear skies along the portion of Route A approaching Sheffield, while the predicted rainfall along routes B and C remains the same.

Accordingly, the risk associated with every possible route between the present location of the vehicle and the destination is recalculated. It is then determined whether route C remains the safest route for the remainder of the journey. After the recalculation, it transpires that due to the decreased rainfall along route A, it is the remainder of route A which is now associated with the lowest aggregated risk coefficient. As a result, a route redirection is proposed by the analysis engine 6.

Subsequently, it is determined whether the risk update related to a risk increase, as previously described. Due to the improvement in driving conditions, it is determined that the risk coefficient update relates to a decrease. As a result, no further action is taken regarding Disturbance 2 and the analysis engine 10 continues to monitor the input devices.

It is important to note that for the purposes of the system and method of the present invention, the source of the data associated with the risk factors, and the information used to derive risk coefficients, is irrelevant. The system and methods of the present invention may function equally well regardless of the actual source of the data.

Example 8.

This example illustrates how a risk calculation for two possible routes - routes A and B - that satisfy a journey specification criteria between a starting location in the Dutch city of Apeldoorn (Anna Bijnsring) and a desired destination in the city of Deventer (Kerkstreet) may be carried out.

The approach is as follows: first the risk values of both routes are calculated, based on risk values of different road types, road crossings (using risk figures from Example 1 , see Table 1 and 4) and the left turn manoeuvre (using risk ratio's from Example 4b, see Table 9). The latter ratio is applied to the average risk at an unsignalized intersection. Then these risk values are made age specific, based on the different risk values per age group per road type (using Example 6 and Figure 14a). We calculate for each age group (based on Figure 14a), the ratio of risk for each road type (referred to as road type X in this explanation) divided by the risk on motorways. We assume that 45 year olds have an average risk and that their risk ratio (risk per road type compared to motorways) is already sufficiently reflected in the risks per road type as described in Table 1. For other age groups we are only interested in the degree to which their risk ratio (risk per road type compared to motorways) is greater than for 45 year olds. Accordingly, the modification factor per age group per road type is estimated as the risk ratio for that age group and road type compared to the risk ratio for that road type for 45 year olds. Suppose for instance that 20 year olds have a 4 times higher risk at rural roads compared to motorways (according to Figure 14a) while this ratio is 3 (according to Figure 14a) for 45 year olds. In this example, a modification factor of 1.33 (4/3) would be assigned to rural roads for 20 year olds. Finally, the contribution to the expected number of accidents for this age group on road type X is estimated as distance travelled on road type X, the base risk of road type X, and a modification factor for road type X for 20 year olds.

Route A is the shortest route; with a length of 15.1 km and a journey time of 22 minutes (around 9PM). Route B uses the motorway as much as possible, has fewer left turns, and is slightly longer (23.2 km / 24 minutes, around 9 o'clock PM). Route A leads the driver through a built-up area for 3.8 kilometres and includes a major rural road over 11.3 kilometres. The route contains a relatively high number of

roundabouts, which have a lower risk value than traditional intersections.

Route B leads the driver through a built-up area for 7.5 kilometres and includes 15.7 kilometres of motorway. A proportion of the intersections have higher risk values than in Route A.

Table 12 shows the results. A lower expected number of accidents indicates a higher level of safety. It is to be appreciated that the calculated values only serve to provide a comparison between the different segments. Thus, for present purposes the most important value is the ratio of expected accidents for route A to the expected number of accidents for route B. A ratio less than 1 indicates that route A is safer for the specific age group, and similarly a ratio greater than 1 indicates that route B is safer for the specific age group. The results indicate that in this example for young and middle aged groups, route A is safest, whereas Route B is substantially safer for older drivers.

Table 12 Estimated number of injury accidents per million vehicle km for two routes between the same origin and destination points

* As Poppe ( 1996) did not report a risk for urban access roads, the risk of single carriageways assigned to this category (to discourage use of this road category)

** As Poppe ( 1996) did not report a risk for rural roundabouts, we used the 70% risk reduction found for rural roundabouts after replacement of rural intersections by Fortuijn (SWOV, 2012). We estimated the risk as the product of (1-0.3) and the 0.22 risk at rural intersections.

*** As the average risk of turning at an intersection is already included in the intersection risk figures, only an additional risk equal to relative risk minus 1 is added to the overall risk estimate. For instance for unsignalized urban, three legs: (2.3 - 1) * 0.092 = 0.120

****For each route the expected number of injury accidents per million km is estimated as sum of the expected accidents per segment and an additional risk for left-turns. The estimation per road type is the product of distance / nr. of intersections, base risk, and age modification factor.

The results indicate that in this example for young drivers both routes don't differ significantly on safety, for young and middle aged groups Route A is the safest and for elderly drivers Route B is the safest.

In yet further embodiments, the methods and navigation systems of the present invention may be adapted to incorporate a risk modification factor accounting for whether any passengers may be located within the vehicle, which could increase the likelihood of the driver being distracted whilst driving within the traffic network. For example, one way in which this may be achieved is by having a plurality of predefined profiles within the driver's user profile, each predefined profile being associated with a different risk modification factor. The driver may then select the predefined profile which is most representative of the driver's current situation. Thus, if the driver is driving in a vehicle with four friends, and therefore there is an increased likelihood of the driver being distracted, a profile associated with an increased risk modification factor may be selected by the driver when inputting the desired destination location in the navigation system. Similarly, if the driver is driving in a vehicle with a plurality of peers, there may again be an increased likelihood of the driver being distracted, and therefore a predetermined profile associated with an increased risk modification factor may be selected when commencing the journey. Similarly, if the driver is commencing a journey late at night, there may be an increased risk of the driver suffering from fatigue, which again may be associated with an increased risk modification factor. The use of predefined user profiles is advantageous in that it provides a convenient and relatively quick means of accounting for specific conditions that the driver may be subject to, which could result in an increase in the risk factor associated with a route.

In yet further embodiments of the invention, the method and navigation system may be adapted to account for driver fatigue resulting from prolonged driving stints. It is widely recognised by traffic and road safety authorities that driver performance levels wane significantly as a function of driving time. The methods and systems of the present invention may account for the increased risk due to fatigue resulting from prolonged periods of continuous driving, by adopting a risk modification factor, whose magnitude is proportional to the period of continuous time that the driver has been driving. For example, the risk modification factor may be adapted to increase as a function of the time the driver has been driving, such that the risk modification factor after an eight hour driving stint will be considerably greater than the risk modification factor associated with a two hour driving stint. This increased risk modification factor may have the effect that routes having, for example, many intersections, and therefore requiring the driver to be particularly alert to potential dangers, may now be associated with greater risk factors than they would otherwise be.

The person skilled in the art will appreciate that the present invention is not limited to the embodiments described herein, which are provided as non-limiting illustrative examples of the invention only. Rather, it is envisaged that numerous changes and modifications may be made to the herein described embodiments without departing from the spirit and scope of the invention as set out in the appended claims, both in the processes described and in the range of applications for which these processes may be used.