It is known to electronically influence the control of a car for improving the active vehicle safety. To this end, the car's control system comprises one or more sensors and a processing unit for estimating the friction of the tires on the road from the signals of those sensors. On the basis of the estimated friction of the tires on the road, the control system derives warning signals, which it brings to the attention of the driver or, even, uses for automatically intervening in the control of the car. An example hereof is the ABS which regulates the wheel slip for utilizing the friction potential between the tire and road surface as much as possible. Another example hereof is the ESP (Electronic Stability Program), which provides for the improvement of the road holding by applying differences between the brake forces on different tires.

In such uses, for the conventional estimation of friction, sensors are utilized to this end, which measure the friction between tire and road surface directly, for instance from the occurring accelerations of the car. These types of direct sensors have the disadvantage that, in fact, the warning signals can only be generated when a significant part of the friction potential has already been used. This can be understood better with reference to Fig. 2, in which, vertically, the friction coefficient has been plotted, as a function of the longitudinal wheel slip. The far right in the Figure corresponds to a freely rolling tire, the far left corresponds to a blocked (non-rolling) tire. Different curves 20a-d show the development of the friction at a water layer thickness of 0. 5, 1,2 and 5 mm, respectively.

The friction between tire and road surface is a non-linear function of the longitudinal wheel slip, after an initial linear behavior for small wheel slip (at the far right in the graph) the friction coefficient, after having reached a

maximum value, decreases from right to left, with increased braking. The friction potential is of importance for determining whether or not braking can still be carried out safely and the wheel slip at which the maximum occurs is of importance for optimal braking. The Figure illustrates that the friction potential and the position of the maximum depend on the water layer thickness, although with small longitudinal wheel slip (far right) the friction between tire and road surface hardly if at all depends on the water layer thickness. This first part is indicated as the so-called linear slip area.

From measurements with a direct sensor on the vertically plotted friction coefficient, the friction potential and the position of the maximum can only be determined if the slip is greater than the area in which the relation between friction coefficient and the wheel slip proceeds linearly. With the direct sensor, this can only be done accurately when a considerable part of the friction potential has been used. Then, it is often too late to correct.

Instead of direct sensors for the friction between tire and road surface, use can be made of indirect sensors. Indirect sensors can predict the effect shown in Fig. 2 before this occurs, and, as a result, generate warnings to take precautions before a dangerous situation arises. Use is then made of a model which indicates how a number of parameters, such as the profile depth of the tires, the macro texture of the road, the water layer thickness et cetera influence the friction between the tire and the road surface. When sensors are available for recording the relevant parameters, the model can predict what the friction potential is and at which longitudinal wheel slip the maximum friction between tire and road surface occurs. Unfortunately, it is difficult and costly to provide. good sensors for all relevant parameters.

It is inter alia an object of the invention to provide a vehicle with a control system which generates warning signals in an improved manner.

The invention provides for a vehicle provided with a control system having

- one or more sensors for measuring interaction parameters between a tire of the vehicle and a road; - an actuation unit; - a calculating unit coupled to the sensors and the actuation unit, and arranged for sending, during driving, an actuation signal to the actuation unit, depending on a predicted friction between tire and road surface of the vehicle, wherein the calculating unit calculates the predicted friction between tire and road surface by determining a value of one or more current parameters of the model from the sensor measurements and a model for the friction between tire and road surface of the vehicle, and by calculating, from the model and the value of the one or more parameters, the predicted friction between tire and road surface for other than current driving conditions.

By using a model for the friction between tire and road surface, it is possible on the one hand, to determine, from measured interaction parameters such as the current friction between tire and road surface, the required parameters of the model (for instance by inverted use of the model), and, on the other hand, with the aid of the determined parameters, to predict the friction between tire and road surface under different driving conditions, in particular, under more extreme driving conditions such as upon increased braking or intensified steering.

Thus, from the measured friction between tire and road surface, under non-slipping conditions, for instance, a prediction of a saturation property of the friction between tire and road surface can be made, such as the friction potential, i. e. the maximum brake force which will be available under the current conditions if braking takes place. This friction potential can, in turn, be used for generating a warning upon speeding, or for an automatic reduction of the speed to a safe value which guarantees a particular maximum length of the brake path.

In one elaboration, the one or more parameters of the model that are estimated from the sensor measurements comprise a value for a

macrostructure parameter of the road, which is determined with the model from measured friction between tire and road surface at the current driving conditions, whereupon the calculating unit, with the thus determined value, determines the predicted friction between tire and road surface. In particular the macrostructure of the road is a parameter which is relevant to the friction between tire and road surface under wet conditions. There are no economically attractive sensors for estimating this macrotexture. By estimating this parameter with the aid of a model of the friction between tire and road surface that a tire gives at different values of the macrostructure parameter, and by then using the estimated value for predicting the friction between tire and road surface under different driving conditions (for instance the maximum value of the friction, or the wheel slip at which this maximum will occur), the safety of the vehicle can be increased in a simple manner. As an alternative, the macrostructure parameter and the microtexture parameter could be determined by the road manager and be radiographically forwarded to the vehicle, but this involves a more complex system which is more susceptible to malfunctions.

If the macrotexture parameter and/or microtexture parameter are not measured, use can be made of an average value for this/these parameter (s).

This average value can also be used for testing the reliability of the estimated macrotexture parameter and/or microtexture parameter.

Preferably, the control system comprises a memory containing information representing the friction between tire and road surface as a function of the following parameters: the speed of the vehicle, the water layer thickness on the road, the parameters specific to the tire which is fitted to the car and the macrostructure parameter-of the road. This enables a simple implementation for calculating a reliable prediction of the friction between tire and road surface. Preferably, the memory also contains information describing the friction between tire and road surface depending on a profile depth of the tire. The profile depth can also be recorded periodically with a profile depth

gauge. As an alternative, the calculating unit can be arranged for estimating the profile depth by averaging values of friction between tire and road surface, which have been determined on the basis of measurements of the sensors at different points in time.

These and other objects and advantages of the vehicle, the control system for the vehicle and the method for controlling the vehicle will be further described with reference to the following figures.

Fig. 1 shows a top plan view of a vehicle ; Fig. 2 shows a graph of a generated brake force; Fig. 3 shows a graph of a blocking value; Fig. 4 shows a graph of a generated transverse force ; Fig. 5 shows a diagram of a control system; Fig. 6 shows a flow chart of the control of the vehicle.

Fig. 1 shows a top plan view of a driving vehicle 10, with tires 12.

Arrow 14 indicates the driving direction, which makes an angle with the longitudinal direction of the vehicle. Moreover, for one of the tires 12 the Figure shows the angle between the respective tire 12 and the driving direction 14. Each of the tires 12 has such an angle. For the sake of perception, the respective angles have been drawn exaggeratingly large.

The invention aims for a control system (not shown in Fig. 1) which generates warning signals with regard to the active safety of the vehicle 10, for bringing these to the attention of the driver and/or for automatically adjusting the control of the vehicle. The control system founds the warnings signals on the basis of a prediction of the friction between tires 12 and road surface upon braking, accelerating and/or steering the vehicle 10. For this friction between tire and road surface, for instance the force the road applies via the tires 12 to the vehicle under various conditions, diverse theoretical and experimental models exist.

The tires 12 serve inter alia for decelerating and accelerating the vehicle 10. A speed difference between the running surface of the tires and the

road surface then generates a brake force or acceleration force Fx, with which the speed of vehicle 10 in the driving direction is decelerated or accelerated.

The brake force can depend on the condition of the tire (profile depth, pressure, temperature etc.), the roughness of the road (grain size, smoothness of grains, presence of moisture, ice, etc.), the speed of the vehicle etc. As a rule, the brake force is proportional to the weight of the vehicle, i. e. the vertical force Fz exerted by the vehicle on the road.

Fig. 2 shows llx (Mu_x), the relation between the generated brake force Fx and Fz on a tire as a function of the ratio K (Kappa) of the rotational speed of the tire and the rotational speed the tire should have for rolling freely at the actual speed of the vehicle 10. K can be expressed in fractions between - 1 and 0 or in corresponding percents. K=-1 (far left) corresponds to a non- revolving (blocking) tire, Ko (far right) corresponds to a freely rolling tire. The Figure shows a number of curves 20a-d for different water layer thicknesses on the road, all at equal speed of the vehicle. From the Figure it will be clear that when K decreases (from right to left), the brake force initially increases proportionally to the change in the ratio K independently of the water layer thickness, but later reaches a maximum depending on the water layer thickness, and then decreases to a blocking value. Parameters such as the maximum and the blocking value additionally depend on the speed of the vehicle.

Fig. 3, for instance, shows curves 30a, b of the blocking value (with K=-1) as a function of the speed for a dry road and a road having thereon a limited water layer thickness, respectively.

The maximum brake force (the friction potential) and the blocking value are relevant parameters for the active vehicle safety. They indicate how quickly the vehicle can be brought to a standstill and are therefore determinative of the length of the brake path and the speed at which one can react to unforeseen circumstances. As, initially, the force Fx is not influenced by the water layer thickness, the position and the magnitude of the maximum

as well as the blocking value cannot simply be predicted from the measured force at a low K value in the linear slip area.

In addition to braking, the tires 12 also serve for determining the driving direction 14 of the vehicle 10. Each tire 12 then applies a transverse force which is dependent on the angle (p between the tire 12 and the driving direction 14. This force increases with the angle (p to a saturation level.

Fig. 4 shows a graph of the generated transverse force Fy as a function of the angle (p. At an increasing angle, the force Fy initially increases, but later the force saturates at a saturation value. Here also, the force depends on the properties of the tire, the road and the driving parameters of the vehicle. For small angles the water layer thickness makes no difference, but in particular the saturation behavior does depend on the water layer thickness and the speed. Accordingly, this saturation behavior cannot be predicted straightaway from the measured forces either. When the centripetal force required for negotiating a bend exceeds the saturation value, the vehicle will become unstable (for instance start slipping). The saturation value is therefore relevant to the safety.

Fig. 5 shows a diagram of a control system for the vehicle of Fig. 1.

The system comprises a number of sensors 20, among which, for instance, a wheel rotation speed sensor, an acceleration sensor, a vehicle rotation sensor and a road condition sensor, such as, for instance, a water layer thickness gauge. Moreover, the system uses, optionally, a sensor with road parameters which are kept up to date by a road manager. The system further contains an actuation unit 28, a calculating unit 26 connecting the sensors 20 and the actuation unit 28, and a memory 29 coupled to the calculating unit 26.

Actuation unit 28 is, for instance, an indicator for giving an indication signal to the driver of vehicle 10, or a control actuator for adjusting the speed of the vehicle 10.

Fig. 6 shows a flow chart of the operation of calculation unit 26. In operation, in a first step 61, the calculating unit 26 receives signals from the

sensors 20. In a second step 62, the calculating unit 26 calculates therefrom one or more parameters that give a prediction of the friction between tire and road surface. In a third step 63, depending on the value of the calculated parameter or parameters, calculating unit 26 sends a control signal to the actuation unit 28.

In an embodiment, calculating unit 26 sends, for instance, a warning signal when the friction potential (the maximum attainable brake power) according to the prediction falls below a threshold value, while the threshold value is, for instance, calculated such that a friction potential above the threshold value guarantees a desired brake path length. In another example, the threshold value is calculated such that the warning signal is generated when according to the prediction, the critical curve speed (the speed above which the curve can no longer be negotiated safely) will be exceeded.

For making the prediction in the second step 62, the calculating unit 26 first carries out a first sub-step 621, in which effectively an estimation is made of the speed, the water layer thickness, the parameters of the tire, a macrostructure parameter of the road (characteristic for the grain size of the grains in the road surface) and, optionally, a microstructure parameter of the road (grain roughness). These parameters are estimated on the basis of a number of signals from the sensors 20. By means of radiographic communication, the micro and macro structure can also be made available from the vehicle to other vehicles. This can be done by means of direct vehicle- vehicle communication or via a central point near the road. In the same manner the vehicle receives information, it can also receive information from the other vehicles with a similar system.

In a second sub-step 622, on the basis of the estimated parameters, the calculating unit 26 predicts the saturation properties of the friction between tire and road surface which will occur under driving conditions other than the current driving conditions. To make the prediction, the control system contains, for instance, tables or empiric comparisons stored in the memory 29

and which contain numerical data of forces as a function of K or a (or saturation values of these functions such as the maximum value as a function of K (or Fz) for a number of values of the following parameters: the speed, the water layer thickness, the parameters of the tire, the macrostructure of the road and, optionally, the microstructure (grain roughness). In effect, therefore, memory 29 stores information which is described in graphs such as those in Figs. 2 to 4.

In this case, in the second sub-step 622, calculating unit 26 uses the estimated parameter values from the first sub-step for retrieving the relevant numerical data from memory 29, to thus be able to predict or calculate the friction between tire and road surface from the tables. Hence, given the estimated parameters, the calculating unit 26 reads the friction potential (the maximum attainable brake force) or, derived therefrom, the speed and/or the minimum following distance from the memory 29.

When numerical data are available for only a limited number of parameter values, calculating-unit. 26 can, if necessary, calculate the forces, the friction potential etc. for the estimated parameter values through interpolation of the numerical data for a number of parameter values for which the numerical data are indeed stored. The calculating unit 26 can also make use of information from memory 29 representing mathematical formulas for the respective forces as a function of the parameters, such as, for instance, coefficients of a polynomial approach.

The numeric data or the respective formulas can, for instance, first be experimentally determined for a number of parameter values or from a theoretic model and then, before the driving of the vehicle, be stored in memory 29. As a rule, the respective numerical data or formulas will depend on the sort of tire which is fitted to the vehicle. Therefore, the numerical data are preferably loaded in memory 29 depending on the type of tire used, or data are stored in the memory for a number of different types of tires. In that case,

later, an identification can be stored of the type of tire fitted to the vehicle, allowing calculating unit 26 to retrieve the relevant data from memory 29.

Even during the lifespan of one tire, the numerical data or formulas for that tire can change, in particular as a function of decreasing profile depth.

In order to take this into account, preferably, memory 29 also stores information describing the forces and/or the friction potential depending on the profile depth. Preferably, calculating unit 26 adjusts the profile depth used, preferably regularly so, for instance on the basis of sensor measurements on the profile depth, and, for the purpose of prediction, uses the data for the profile depth used.

As to the sensors 20, wheel speed sensors, accelerating sensors and vehicle rotation sensors are naturally generally known. On the basis of measurements of such sensors, in a simple manner, the occurring forces Fx, Fy and the slip values K and a can be determined for the current driving conditions. Here, an estimation algorithm is used that calculates the slip values K and a and the occurring forces Fx and Fy for the front axle and rear axle separately. The model is a so-called one track model in which the two front wheels and the two rear wheels, respectively, are viewed as one tire. For measuring the water layer thickness, for instance, use can be made of a road surface condition sensor which can monitor the water layer thickness in an optical way.

For measuring the macrostructure parameter of the road surface, for instance, use can be made of a laser sensor describing the form of the macro texture. This complex measuring method cannot be used on a vehicle produced in series and is reserved for specialized testing institutes. This or a different sort of measurement of the macro structure parameter can, optionally, be carried out outside the vehicle, for instance by a road manager and be transmitted wirelessly to calculating unit 26, for instance by electromagnetic signals, acoustic signals, modulated light etc.

However, this type of solution is, as far as can be judged now, economically unattractive. Instead of explicitly measuring the macrostructure parameter, the calculating unit 26 therefore preferably calculates the macrostructure parameter on the basis of the measured forces and known parameters of the tires. Calculating unit 26 does this through a calculation of the macrostructure parameter from the measured forces Fx and/or Fy at the measured K and/or a, for instance by comparing these measured forces with the predicted forces for a number of values of the macrostructure parameter and by selecting therefrom the macrostructure parameter which predicts the measured Fx and/or Fy. Thus, on the basis of the current driving conditions, an estimation is made available which is used to predict the grip on the current road under different (more extreme) driving conditions.

For determining the macrostructure parameter, the calculating unit 26 preferably uses the longitudinal rigidity Kx and/or the transverse rigidity Ky. Ky is defined as the ratio between the derivative of the generated force Fy and the angle a: Ky=dFy/da. For small angles a occurring under non-extreme driving conditions, this transverse rigidity is constant as a function of the angle a depending on the macrostructure and virtually independent of the speed. The longitudinal rigidity Kx is defined as the ratio between the derivative of the generated force Fx and K : Kx-dFx/dK. For K small values occurring under non-extreme conditions, this longitudinal rigidity is constant as a function of K depending on the macrostructure and virtually independent of the speed. Therefore, by determining Kx and/or Ky from the measured forces and K and/or a, while driving under non-extreme conditions, the macrostructure can be determined.

The profile depth too can, if desired, be estimated from measurements of the forces Fx and/or Fy at differently measured K and/or a.

This is, for instance, possible when the macrostructure parameter is known at least for a number of different road parts. A measured rigidity I<x and/or Ky corresponds to a set of possible combinations of the profile depth and the

macrostructure parameter. When the macrostructure parameter of a particular part of the road is known, the profile depth can thus be determined directly from the measured Kx or Ky. Such a parameter can be transmitted to the vehicle, for instance radiographically, when the vehicle passes the respective part of the road.

Even when the macrostructure is not known anywhere, the profile depth can be determined if it may be assumed that, over a longer period of time, the vehicle will drive over roads of varying macrostructure. In this case, by using Bayesian estimation techniques, an estimation of the profile depth can be obtained by weighting the possible profile depths at measured Kx and/or Ky values with the probability of the associated macrostructure parameter and by averaging the thus obtained, weighted profile depth over a longer period of time (the probability of the associated macrostructure parameter can here for instance be determined on the basis of known probabilities for the roads over which the vehicle drives according to navigation data).