WO2011054363A1

WO2011054363A1 - Surface classification

Info

Publication number: WO2011054363A1
Application number: PCT/EP2009/007912
Authority: WO
Inventors: Fredrik Gustafsson; Marius RÖLLAND; Rickard Karlsson; Andreas Hall
Original assignee: Nira Dynamics Ab
Priority date: 2009-11-04
Filing date: 2009-11-04
Publication date: 2011-05-12
Also published as: DE112009005342T5; DE112009005342B4

Abstract

A method for estimating the ground condition under a driving vehicle having at least one pair of a first and a second wheel, the method comprising for at least one pair of wheels: determining first and second sensor signals indicative of time dependent vibrations at the front and rear wheels, respectively; correlating the first and second sensor signals in order to determine a correlation signal indicative of the time delay between the first and second sensor signals; determining the ground condition based on characteristics of the correlation signal.

Description

SURFACE CLASSIFICATION

FIELD OF THE INVENTION

The present invention relates generally to the estimation of the road surface condition under a vehicle and, for example, to systems, methods, and computer program products for estimating the road surface condition under a vehicle.

BACKGROUND OF THE INVENTION

Modern cars comprise electronic control systems like anti-lock-braking systems (ABS), dynamic stability systems, anti-spin systems and traction control systems. Besides these active control systems there also exist driver safety information systems such as road friction indicators and sensor-free tire pressure monitoring systems which present information about the driving condition to the driver.

All the above-mentioned systems benefit from the knowledge about the road surface condition under the vehicle. Several different techniques are used in the prior art to determine the road surface condition under a driving vehicle. One such technique is based on vertical accelerometers in a suspension system of a car. Another technique is based on level meters in the fuel tank of the car. Other techniques use special air mass flow sensors in the engine control unit.

The present invention relates to techniques for estimating the road condition which make use of the vibration signals obtained from sensors, e.g. wheel speed sensors or acceleration sensors. For example, using the signals from wheel speed sensors of ABS systems (and/or from the vehicle's internal CAN-bus) provides an economical way to road surface condition measurements since these ABS systems belong to the standard equipment of the majority of the cars and trucks sold today.

Such a system which is based on the signals of wheel speed sensors is for example disclosed in US-patent 5,566,090 which is directed to a method for detecting segments of bad road directly from the raw data provided by an ABS sensor. The method uses the fact that segments of bad road result in strong fluctuations of the wheel speeds of the car. Strong wheel speed fluctuations in turn result in large differences between successive segment times, where the segment time is the time the wheel needs to pass through associated angle segments. The disclosed method determines a segment of bad road if the difference between successive segment times is greater than a pre-set limit value. This simple decision algorithm operates directly on the raw signals of the wheel speed sensor. The US 4,837,727 discloses a method which is based on a similar decision algorithm.

Another system which is based on the signals of wheel speed sensors is disclosed in published international application WO 2005/068271. This system uses a signal correction section to determine an imperfection-corrected sensor signal.

EP 0 795 448 A2 discloses a road surface condition detection system which comprises a wheel speed sensor for detecting a wheel speed of at least one wheel to generate a wheel speed signal and a control unit which integrates the wheel speed signal for a predetermined period of time. The control unit determines a rough road surface condition when the integrated signal is above a predetermined threshold value and, otherwise, a normal road surface condition. Before the integration, the wheel speed signal is band-pass filtered in the frequency range of 10-15 Hz.

It is also known in the prior art that a correlation analysis of vibrations in front and rear wheel speed signals can be used to determine the absolute velocity of a vehicle. By determining the time delay τ between the disturbances of the front and rear wheels the absolute velocity v of the vehicle can be calculated from the relation r =B/v where B is the spacing between the axles of the front and the rear wheel. Such a velocity determination by means of a correlation analysis is e.g. disclosed in published international patent application WO 2005/005993 A1.

SUMMARY OF THE INVENTION

The invention is directed to methods, systems and computer program products for estimating the ground condition under a driving vehicle as defined in the appended claims. According to a first aspect, a method is presented for estimating the ground condition under a driving vehicle having at least one pair of a first and a second wheel, the method comprising determining first and second sensor signals indicative of time dependent vibrations at the first and the second wheel, respectively; correlating the first and second sensor signals in order to determine a correlation signal of the first and second sensor signals; and determining the ground condition based on characteristics of the correlation signal.

According to a further aspect, a system is presented for estimating the ground condition under a driving vehicle having at least one pair of a first and a second wheel, the system comprising: first and second sensors determining first and second sensor signals indicative of time dependent vibrations at the first and the second wheel, respectively; a correlation unit for correlating the first and second sensor signals in order to determine a correlation signal of the first and second sensor signals; an evaluation unit for determining the ground condition based on characteristics of the correlation signal.

According to another aspect, a computer program product is presented adapted for estimating the ground condition under a driving vehicle having at least one pair of a first and a second wheel, the computer program being arranged for: determining first and second sensor signals indicative of time dependent vibrations at the first and the second wheel, respectively; correlating the first and second sensor signals in order to determine a correlation signal of the first and second sensor signals; determining the ground condition based on characteristics of the correlation signal.

Other features are inherent in the methods and systems disclosed or will become apparent to those skilled in the art from the following detailed description of embodiments and its accompanying drawings.

DESCRIPTION OF THE DRAWINGS

Embodiments of the invention will now be described by way of example and with reference to the accompanying drawings, in which: Fig. 1 shows a car driving on a road with a surface disturbance ("bump") and the time dependent rear and front wheel noise signals resulting when the car passes this surface disturbance;

Fig. 2 is a schematic view of an embodiment of a wheel speed sensor;

5 Fig. 3 shows an exemplary diagram of four wheel speed signals obtained from the four wheels of a driving vehicle as a function of time;

Fig. 4 shows the correlation function of the signals in Fig. 3;

Fig. 5a schematically displays the correlation signal in the correlation function for a rough road condition; the dashed line indicates the envelope of the i o correlation signal;

Fig. 5b schematically displays the correlation signal in the correlation function for a smooth road condition; the dashed line indicates the envelope of the correlation signal;

Fig. 5c schematically displays how fluctuations in the correlation signal can result i s in side maxima being stronger than the main maximum of the correlation signal; the dashed line indicates the envelope of an ideal correlation signal which is not influenced by fluctuations;

Fig. 6 schematically describes the Full Width Half Maximum (FWHM) technique which is used in an embodiment to determine the width of the correlation 20 peak;

Fig. 7 shows the correlation function of the correlation between data from the front and the rear wheel in the event domain. At the delay of 67 cogs the maximum peak value is located;

Fig. 8 shows a plot of the delay which corresponds to the maximum of the 25 correlation signal for a large number of batches of measurement data.

Each circle represents the maximum correlation value (correlation peak) from a sample of the monitored sensor signal; Fig. 9 shows an exemplary embodiment of a method to determine the ground condition under a driving vehicle;

Fig. 10 shows a further exemplary embodiment of a method to determine the ground condition under a driving vehicle.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

In general the presented method and system for determining the road surface condition under a vehicle is based on a correlation analysis of the time dependent behavior of vibration signals, e.g. vibrations in wheel speed signals, of a vehicle's pair of wheels which vary with road bumpiness or unevenness. The cause for such variations may be any small or large road feature like asphalt texture, splits, small stones, bumps, etc. These features induce these vibrations via the tire-road contact. The variations which are induced at a particular wheel result in a specific time dependent behavior of the respective sensor signals. If wheel speed sensors are used to measure the vibrations, then the wheel speed signals are preferably measured in a pair of a front and a rear wheel which are running in a lane so that they feel the same road features in a time delayed manner. The instantaneous wheel speed signals measured by these sensors are influenced by the road features which were instantaneously passed by an individual wheel.

Fig. 1 schematically shows a car with four wheels which is driving on a road with a surface disturbance ("bump"). The two graphs below the schematic representation of the car display the time dependent rear and front vibration signals resulting when the car passes the surface disturbance. It can be seen in the two graphs that the signal resulting from the surface disturbance appears in the rear wheel vibration signal with a time delay τ compared to the front wheel vibration signal.

The vehicle may be any wheeled vehicle, like cars, lorries, trucks, motorcycles, trains, etc. which have a front and a rear wheel in contact with ground. In the preferred embodiments, the two wheels which are running in a lane (a front wheel and a rear wheel) are referred to as belonging to a particular pair of wheels. In general, the front and the rear wheel are mounted on different axles but the axles are not required to be the first and the last axle of the vehicle. For example, the front-left and the rear-left wheel of a four wheeled car may constitute an appropriate wheel pair. In vehicles with more then two axles, two arbitrary axles of the entire set of axles may be chosen as the two wheels in a lane which are here denoted as front and rear wheel. In some embodiments, the front and the rear wheel run in a lane so that during straight driving road features which are passed by the front wheel are subsequently passed by the rear wheel. Wheel speed variations at the two wheels of such a pair of wheels are the basis for the velocity determination as presented herein. More then one of the above defined wheel pairs may be included in a velocity analysis to enhance the performance of the system, but, in the following embodiments, the principles of the velocity determination method are presented with one pair of wheels only. The wheel pair must not necessarily be a pair of front and rear wheels. In other embodiments, a pair of a left wheel and a right wheel are used.

The two axles respectively the two wheels are spaced by a distance which is in the following denoted as wheel spacing. Commonly, this wheel spacing β is also denoted as the wheel base.

The sensors used to obtain the front and real wheel noise signals may be of any type which is responsive to vibrations resulting from the contact of the vehicles' front and rear wheels with the road surface. For example, the sensors may be any common wheel speed sensors. However, any other sensor type may also be used, e.g. accelerometers in a suspension system of a car, ultrasound sensors, microphones, laser sensors, axel height sensors, any other analog distance sensors, geophones which convert displacements into voltage, or e.g. in-tire pressure/accelerometer sensors.

Preferably, the wheel speed sensors of an antilocking system (ABS) are used in the embodiments since such ABS-sensors are already mounted in a majority of the vehicles today. Wheel speed sensors are well known to the person skilled in the art. Fig. 2 shows a schematic diagram of a wheel speed sensor comprising a toothed wheel with seven identical teeth. A sensor is located at the circumference of the toothed wheel. The sensor is arranged to generate a sensor signal whenever a tooth (cog) of the toothed wheel passes the sensor. The sensor may be an optical sensor, a magnetic sensor (e.g. a HALL sensor) or any other conceivable type of sensor. The sensor produces electrical signals which are transported by wires or radio transmission to a subsequent unit for further processing. In the example of Fig. 2, there are in total seven sensor signals generated during one complete revolution of the toothed wheel.

In more detail, the sensor of the wheel speed sensor may internally generate a signal with two possible states, high and low (e.g., high indicating a covered sensor and low indicating an uncovered sensor), which in turn triggers the output of a clock signal delivered from a timer unit (not shown), and outputs a data stream. The data stream comprises data samples in form of, for instance, a real or integer number t(n) which is representative of the time instance of the occurrence of a corresponding internal signal. The time span t(n) = t{n) - t(n-1) is defined as the duration of time between two successive internal signals. Thereby, n is an integer number which denotes the sample number, i.e. n = 1 corresponds to the first sensor signal, n = 2 to the second sensor signal, etc. The sequence t(n) can be transformed from the event domain to the time domain by known means to obtain a wheel speed signal as function of time co(f).

The principles underlying the present invention are in the following explained in the time domain. However, a wheel speed sensor as described above typically provides its measurement results in the event domain. Wheel speed sensors thus do not provide their data in a way which allows a direct storage in time domain. Consequently, when implementing the invention, an analysis in the event domain might be preferable.

As described above, a wheel speed sensor triggers a signal each time the observed wheel has rotated by a particular angle (a = 2%IL in the above embodiment, where L is the number of cogs on the toothed wheel). Consequently, the signal values (angles) are here equidistantly distributed over the signal axes, whereas the corresponding time instances are not equidistantly distributed over the time axes. The time intervals between two sensor signals depend on the rotational velocity of the observed wheel. Such type of data which is for example generated by sampling the output from a wheel speed sensor is generally referred to as event domain sampled (here angle domain sampled; in some cases called "cog" domain). Even though an analysis in the event domain might be preferable, the principles described in this application are equally applicable in the time domain and in the event domain. The skilled person can adapt the methods to the preferred domain by conventional means which are e.g. addressed in the published international patent application WO 2005/005993 A1 already addressed above.

Here, the principles of the road surface classification are explained with reference to a continuous time angular wheel speed signal ω(ή. With 'continuous time' signal a real or hypothetical signal is denoted which provides a data point for every time instance in the available time interval. Some sensors, however, provide sampled measurement signals which are not continuous time signals but discrete signals since, for example, in digital measurements only a limited number of measurement values is recorded.

Let ω{ΐ) denote the velocity of each wheel / = 1 ,2, ... of a vehicle. The embodiment which is shown in Fig. 1 applies to a four-wheeled car where the front- left wheel is numbered 1 , the front-right 2, the rear-left 3 and the rear-right 4. Examples of wheel speed signals sampled by the wheel speed sensors are shown in Fig. 3. The plot shows measured wheel speeds in rad/s of the four wheels of a car as function of time. The data was recorded over a time interval of 60 seconds. The wheel speeds are shown in the interval from 41.9 rad/s to 42.7 rad/s which illustrates that the fluctuations of the wheel speeds around their median are in the range of some percent. In Fig. 3, 40 rad/s corresponds roughly to 40 km/h which is about 11 m/s. Plotted are four curves corresponding to the four wheels FL = front-left, FR = front-right, RL = rear-left, RR = rear-right of the car. One curve contains an offset that is due to different tire radii, wheel slip or cornering. Both curves contain a barely visible disturbance that is delayed by 0.3 seconds.

In preferred embodiments the sensor signals contain a road induced disturbance which appears in the rear axle signals fi¾ff) and ω ) τ seconds later than in the front axle signals o (t) and c^i)- A correlation analysis of the front and the rear sensor signals shows a specific correlation signal, like one or more peaks of the correlation function, which is indicative of this time delay τ between the front wheel and rear wheel speed signals.

In an embodiment, the cross correlation between the front and rear wheel speeds is defined as

*₁₃ (r)

(/ - τ) - E(<^y ₃ (ί - τ)))] ( 1 ) with Ε(ω(ή) denoting the expectation value of ω( . Small disturbances injected by uneven road surface will occur first on o>_\(f) and B/v seconds later on Here v denotes the velocity of the left side of the car and B is the wheel base of the vehicle. This cross correlation function is a function of time delay τ and will show a peak at τ = B/v. It should, however, be noted that the correlation function must not necessarily be a function of time. For example, in an alternative embodiment, the correlation function might be represented in the cog domain. Or, in other embodiments, it might be represented in frequency space in which case the spectrum of the road surface vibration would be reflected in the amplitude of the phase function /¾(/)/ ¾( ) of the Fourier transforms ¾(/) and i¾( ) of the wheel speed signals {t) and β¾(ί). Correlation analysis in frequency space is described in more detail in published international application WO 2005/005993 A1 .

The plot of Fig. 4 shows the correlation function

of the wheel speed signals of the front-left and the rear-left wheel and the correlation function /¾4(*) of the wheel speed signals of the front-right and the rear-right wheel of a car. The abscissa of the plot represents the time delay τ in the interval from -1 to 1 seconds. The ordinate denotes the corresponding values of the cross correlation function Ri₃(i). Each correlation function shows a correlation signal which is indicative of the time delay between the front and rear wheel vibrations, e.g. as here a peak of maximal oscillation for a particular time delay τ. These peaks indicate that disturbances in the wheel speed signals are most similar at these particular time delay values.

The plot of Fig. 4 also shows that the correlation signal develops an oscillatory characteristic with a main lobe and a corresponding main maximum (and a main minimum) in the center of the correlation signal and several side lobes with corresponding side maxima (and side minima) in the vicinity of the main maximum. In the following, the main maximum and the main minimum are considered as belonging to the main lobe of the correlation signal and the side minima and side maxima are considered as belonging to side lobes of the correlation signal. The invention is based on the observation that different road surfaces will excite the cross correlation differently and thus produce a correlation signal which can be used for estimating the road surface condition under a driving vehicle, i.e. for road surface classification. Typically, the main correlation peak (or main lobe) of the correlation signal is weaker developed when a vehicle is driving on a smooth surface. Correspondingly, the side lobes of the correlation signal are weaker developed when a vehicle is driving on a rough surface. The rougher the road is, the less pronounced are the side lobes. An explanation for this effect is given in the following:

There are in general three contributions to the measured sensor signal y,(f):

• a first contribution g * v{f) which originates from the wheel revolutions and which thus reflects the dependency on the vehicle velocity; here v{f) is the wheel speed and g is the impulse response from wheel speed to vibrations; * denotes the convolution operation;

• a second contribution h * z{t) which has its origin in the vertical deflections of the road under wheel /^'; here z,{t) denotes the vertical deflections and h is the impulse response from road deflections to vibrations; and

• a third contribution e,{f) which stems from measurement noise at sensor /.

The measured sensor signal y{f) (in a preferred embodiment the signal ω,(ή measured by a wheel speed sensor) can thus be expressed as follows: y, (t) = g * v_i(t) + h * z_i (t) + e_i (t) (2)

The cross correlation function (equation (1 )) of the sensor signal can then be written as

R_yjj (?) = E(y, (t)y_j (t - r)) = h * h * R_{2 lj} (r) (3)

Here, h (t) = h(-t) denotes the time reversed impulse response and r?_Z|/y(r) the cross correlation of the vertical deflections z{f). As can be seen from equation (3), the contribution from the velocity dependence v,-(f) and the contribution from sensor noise e,{f) disappear. Since h corresponds essentially to the wheel suspension dynamics, h x h resembles a non-causal band-pass filter which is characterized by one main lobe and several smaller side lobes.

Smooth surfaces tend to have a white deflection z(f) so that R_{z j}{ is an impulse function: Rz _jir) = δ(ΒΙν-τ). Here, ^denotes the Dirac delta function, B the wheel base and v the vehicle velocity. The convolution with δΐβΐν-τ) reflects a shift of h * h (i.e. of the main lobe and side lobes) on the time axis. In the specific case that the wheels are on the same axle, i.e. B = 0 and R_z,ij{ = the expected correlation result R_y,ij(r) is h * h , i.e. a symmetric function with main lobe and main maximum at τ= 0.

On rough roads, however, R_{2 j}{r) has a band-pass character. This results in that the cross-correlation of the sensor signal ¾_,/,{ r) is a smoothed version of h * h . The side lobes are less pronounced on rough roads.

The above analysis shows that it is not essential that the sensor signals of a front and a rear wheel are correlated with each other. The principles may also be applied, for example, by correlating sensor signals associated with left and right wheels. Thus, in alternative embodiments, the sensors of a left and a right wheel of the same vehicle axis are correlated with each other. However, in the preferred embodiments described here in more detail, sensor signals of front and rear wheels are correlated with each other.

Figs. 5a and 5b show two correlation signals for two different ground conditions, namely Fig. 5a the correlation signal of a vehicle driving on a rough road surface and Fig. 5b the correlation signal of a vehicle driving on a smooth road surface. In the rough road case of Fig. 5a, the correlation signal has a pronounced main lobe maximum and significantly smaller side lobe maxima. In the smooth road case shown in Fig. 5b, the correlation signal has side lobe maxima which are nearly as pronounced as the maximum of the main lobe. This difference can also be seen in the envelope curves of the correlation function which are shown as dashed lines in Figs. 5a and 5b. The envelope curve in Fig. 5a raises and falls steeper than the envelope curve in Fig. 5b. In other words, the correlation signal in Fig. 5b is broader in shape than the correlation signal in Fig. 5a. This shows that characteristics of the correlation signal, e.g. the shape of the correlation signal which reflect the development of the side lobes depend on the ground condition under a driving vehicle and can thus be used for road surface classification.

As shown above, there are several ways to analyze characteristics of the correlation signal in order to determine the ground condition under a driving vehicle:

For example, in one embodiment, the height of the main maximum is related to the heights of the side maxima. Typically, the ratio between the height of the main maximum and the heights of the side maxima is larger for rough road surfaces than for smooth road surfaces.

According to another embodiment, the envelope of the correlation function as displayed by the dashed lines in Figs. 5a and 5b is determined for each correlation signal by conventional means, e.g. by applying a corresponding filter to the correlation function. The width of each correlation signal is determined by computing the width of each envelope. The width of the envelope can for example be determined by the conventional Full Width Half Maximum technique (FWHM), which is schematically described in Fig. 6, or any other suitable means for quantizing the width of the envelope. The width of the correlation signal can then be used to deduce the ground condition under the driving vehicle.

According to yet another embodiment, the envelope of the correlation signal is computed as in the previous embodiment; but instead of the width of the correlation signal, the surface under the envelope function is used as characteristic for determining the ground condition under the driving vehicle. The surface under the envelope of the correlation signal can for example be calculated by integration.

The skilled person can of course use other ways to quantize characteristics of the correlation signal to determine the ground condition under the driving vehicle, e.g. by applying filters of various types, e.g. Kalman filters or Least Mean Squares autoregressive filters. For example any shape factors as disclosed in published international application WO 2008/113384 might be used to quantize the characteristics of the correlation signal.

In a further embodiment, the front and rear wheel sensor signals are collected during specific time intervals. The sensor signals collected in a time interval form a batch of measurement signals. Such a batch of measurement signals may e.g. have a length of 60 seconds as shown in the example of Fig. 3. The length of a batch might as well be smaller, e.g. 1 second, or larger than the 60 seconds displayed in Fig. 3. In this embodiment, an individual correlation function is calculated for each of the measurement batches and a specific feature of the correlation signal, e.g. the time delay f which corresponds to the maximum of the correlation function is determined for each batch. The time delay which corresponds to the maximum of the correlation function is e.g. obtained from R (r) by = arg max i?_]3 (T) ^

In the case of a vehicle driving on a rough road surface, the main lobe is strong and the side lobes are weak. There is thus a high likelihood that the determined maximum of the correlation function is the central peak of the correlation signal, i.e. the peak which is associated with the main lobe of the correlation signal as displayed in Fig. 5a. The thus determined time delay f should correspond to the expected time delay r = B/v where β is the wheel spacing and v is the vehicle velocity. Due to fluctuations in the measurement signal, there is however a small likelihood that a side lobe maximum of the correlation signal is more pronounced than the central maximum of the main lobe of the correlation signal. This situation is depicted schematically in Fig. 5c. In this case the global maximum of the correlation function determined according to equation (4) is situated in a side lobe of the correlation signal, approximately 10% off from the expected value B/v. The likelihood that the global maximum of the correlation signal is a side lobe maximum is larger when the side lobes are pronounced, i.e. under smooth road conditions. Under rough road conditions, the side lobes are less pronounced and the likelihood that the global maximum of the correlation function is a side lobe maximum is smaller.

Fig. 7 shows an example result obtained from correlating data from the front and the rear wheel in the cog domain. At the delay of = 67 cogs the maximum peak value is located.

Fig. 8 shows a plot of the correlation results of a large number of measurement batches. The plot was obtained from an analysis performed in the cog domain. Each circle represents the maximum correlation value (correlation peak) of one batch of the monitored sensor signal, e.g. the correlation peak of a 60 seconds batch of the monitored sensor signal as shown in Fig. 3. On the y-axis is plotted the correlation delay ή in the cog domain (which corresponds to the correlation delay in the time domain), determined by the correlation analysis, and on the x-axis the time of gathering of the batch data. As time evolves with advancing vehicle position, the circles move from left to right on the x-axis. Three different roads with different road coarseness are displayed in Fig. 8. As described above, with a smoother road surface (road segment 3 in Fig. 8), the expected side lobes are stronger so that the likelihood that a correlation peak (the global maximum of the correlation function of the batch data) will fall into one of the side lobes is larger. This means that for smoother roads, the number of circles which are close to the expected value of the main lobe maximum tends to decrease, and the number of circles which are close to the expected side lobe maxima tends to increase. On the other hand, with a rougher road surface (surface segment 1 in Fig. 8), the expected side lobes are weaker so that the likelihood that a determined correlation peak will fall into one of the side lobes is smaller. This means that for rougher roads, the number of circles which are close to the expected value . of the main lobe maximum tends to increase, and the number of circles which are close to the expected side lobe maxima tends to decrease. In the cog domain, the expected delay would be η = B L 1{2πή, where 8 is the wheel base, L is the number of cogs on the wheel and r is the wheel radius. In a time domain analysis, the expected delay would be r = B/v.

As a result, the circles plotted in Fig; 8 show a band structure with a main band and side bands. The main band reflects the main lobe of the correlation function and the side bands reflect the side lobes of the correlation function. The more circles there are in the side bands, the smoother the road is. The more circles there are in the main band, the rougher the road is. The center of the main band is at η - 65, and the outer bands are centered around η - 59 and η = 71. As the plot of Fig. 8 was obtained in an event domain analysis, the values 65, 59 and 71 are related to the cogs of a wheel speed sensor.

As can be seen from the above, the statistical dispersion of the measured correlation delay (respectively^ ) e.g. the variance

Var ) = E((i -E(i ))²) (5) of the position of the correlation peak can be used as a measure for the ground condition under the driving vehicle. The variance of the positions of the correlation peaks is small for road surface 1 (rough road) in which the main band is strongly filled with circles and the side bands are only weakly filled. The variance of the positions of the correlation peaks gets larger for road surface 2 in which the occupation of the main band is diminished and the side bands get stronger occupied with circles. Finally, the. variance of the positions of the correlation peaks gets largest for road surface 3 (smooth road) in which the occupation of the main band is even more diminished and the side bands get even stronger occupied by the circles.

Fig. 9 shows a flow-diagram which demonstrates the basic steps of an exemplary method to determine the ground condition under a driving vehicle as explained above, in steps 91 and 93, front and rear wheel speed signals are obtained from wheel speed sensors of the vehicle. In step 95, these wheel speed signals are correlated with each other, e.g. by a correlation function R13 according to equation (1). In step 97, the characteristics of the correlation function, respectively the characteristics of a correlation signal, e.g. of a correlation maximum are determined. This may be done on the principles as described above, e.g. by evaluating the strength of the side lobes, e.g. by comparing the strength of side lobes to the strength of the main lobe (main maximum). Finally, in step 99, the ground condition can be determined from these characteristics of the correlation signal. If for example, the shape of the correlation signal is evaluated, e.g. the breadth of the correlation signal, then the broader the correlation signal is, the smoother the road surface is.

As an alternative to the method of equation (5), the skilled person can imagine several other ways of determining a measure for the variance of the correlation delay, i.e. the variance in the position of the correlation peaks. For example, each circle may be attributed to either the expected main band or to an expected side band of the correlation signal. The number of circles falling in the side band of the correlation signal may then be compared with the number of circles falling in the main band of the correlation signal in order to obtain a measure for the road surface condition. This alternative is depicted in Fig. 10. In steps 101 and 103, a total number of N batches of wheel speed signals are accumulated over time. In step 105, for each batch, the correlation of the front and wheel speed signals is calculated. In step 107, the maximum of each correlation function is determined. In step 111 , the total number N1 of batches whose maximum falls into the main lobe is counted. In step 113, the total number N2 of batches whose maximum falls into side lobes is counted. In step 113, the road condition is determined from the ratio of N1 and N2. The larger the ratio N1/N2 is, the rougher the road is.

In general, a correlation analysis of a measurement batch as described above provides one or more output values which reflect e.g. the (current) strength of the side lobes of the correlation function. As said, this output value may for example be the variance of the correlation peak. A real-time road classification can be based on these output values. For example the variance may be approximated in real-time by a Least-Mean-Square filter:

Here, η_Ιι is the raw correlation peak, ή_Ιί is the tracked correlation peak, k is the current batch number, and the step length μ_η is a filter parameter. The classification of the road surface is updated based on the new, filtered value from the correlation analysis. The variance may further be estimated with a combination of a Least-Mean- Square filter and a forgetting factor according to:

Var_k = 0.5«(77, - )² + (1 - u)Var_k_, (7)

Here, u is the forgetting factor of the filter.

A classification of the road condition based on the filtered output value may then be obtained by calibration, i.e. the output results are related to known road conditions in test drives.

Also, a sequential analysis technique might be used to evaluate the output of the correlation analysis, e.g. a cumulative summation technique (CUSUM) may be used to detect changes in the output values.

Hysteresis function may be used to prevent fast fluctuation of the classification results in situation where the surface changes smoothly.

The embodiments of the computer program products with program code for performing the described methods include any machine-readable medium that is capable of storing or encoding the program code. The term "machine-readable medium" shall accordingly be taken to include, but not to be limited to, solid state memories, optical and magnetic storage media, and carrier wave signals. The program code may be machine code or another code which can be converted into machine code by compilation and/or interpretation, such as source code in a high- level programming language, such as C++, or in any other suitable imperative or functional programming language, or virtual-machine code. The computer program product may comprise a data carrier provided with the program code or other means devised to control or direct a data processing apparatus to perform the method in accordance with the description. A data processing apparatus running the method typically includes a central processing unit, data storage means and an l/O-interface for signals or parameter values.

Thus, a general purpose of the disclosed embodiments is to provide improved methods and products which enable to more accurately determine a ground condition by means of sensors located at first and second wheels of a vehicle.

All publications and existing systems mentioned in this specification are herein incorporated by reference.

Although certain methods and products constructed in accordance with the teachings of the invention have been described herein, the scope of coverage of this patent is not limited thereto. On the contrary, this patent covers all embodiments of the teachings of the invention fairly falling within the scope of the appended claims either literally or under the doctrine of equivalents.

Claims

CLAIMS . A method for estimating the ground condition under a driving vehicle having at least one pair of a first and a second wheel, the method comprising:

a) determining first and second sensor signals (ω) indicative of time dependent vibrations at the first and the second wheel, respectively;

b) correlating the first and second sensor signals (ω) in order to determine a correlation signal of the first and second sensor signals {ω)

c) determining the ground condition based on characteristics of the correlation signal.

2. The method of claim 1 in which the first wheel is a front wheel and the second wheel is a rear wheel of the vehicle.

3. The method of claim 1 or claim 2 in which the step of determining the ground condition is based on evaluating the shape of the correlation signal.

4. The method of anyone of the preceding claims in which

the correlation signal comprises a main lobe and side lobes, and

the step of determining the ground condition comprises evaluating the strength of the side lobes of the correlation signal.

5. The method of claim 4 in which evaluating the strength of the side lobes of the correlation signal comprises comparing the strength of the side lobes to the strength of the main lobe of the correlation signal.

6. The method of anyone of the preceding claims in which

in step a) the first and second sensor signals are determined for multiple batches of sensor signals;

step b) comprises correlating the first and second sensor signals for each of the batches and determining the position of the maximum of the correlation signal for each of the measurement batches;

step c) comprises determining the statistical dispersion (Var( f )) of the position of the maximum of the correlation signal;

7. The method of claim 6 as far as dependent on claim 5 in which

comparing the strength of the correlation signal in the main lobe to the strength of the correlation signal in the side lobes comprises

for. each of the batches, associating the position of the maximum of the correlation signal to either the main lobe or to the side lobes of the correlation signal;

determining the number of batches whose maximum of the correlation signal is associated with the main lobe;

determining the number of batches whose maximum of the correlation signal is associated with the side lobes;

relating the number of batches whose maximum of the correlation signal is associated with the main lobe to the number of batches whose maximum of the correlation signal is associated with the side lobes;

8. The method of anyone of the preceding claims, further comprising filtering the correlation result so that the characteristics of the correlation signal are obtained.

9. The method of claim 8, in which the filtering step comprises determining the envelope of the correlation signal.

10. The method of anyone of the preceding claims in which the sensor signals are wheel speed signals {ω) indicative of the time dependent behavior of the first and second wheel speeds of the vehicle.

1 1. The method of anyone of the preceding claims in which the sensor signals are signals from accelerometers, located e.g. in a suspension system of the vehicle.

12. The method of anyone of the preceding claims in which the classification of the road surface is updated based on the new, filtered value from the correlation analysis.

13. System for estimating the ground condition under a driving vehicle having at least one pair of a first and a second wheel, the system comprising: a) first and second sensors determining first and second sensor signals (ω) indicative of time dependent vibrations at the first and the second wheel, respectively; b) a correlation unit for correlating the first and second sensor signals (ω) in order to determine a correlation signal of the first and second sensor signals (a>); c) an evaluation unit for determining the ground condition based on

characteristics of the correlation signal.

14. Computer program product adapted for estimating the ground condition under a driving vehicle having at least one pair of a first and a second wheel, the computer program being arranged for:

a) determining first and second sensor signals (a>) indicative of time dependent vibrations at the first and the second wheel, respectively;

b) correlating the first and second sensor signals (a>) in order to determine a correlation signal of the first and second sensor signals (a>);

15. Computer program product according to claim 14, further arranged for evaluating the method steps of anyone of claims 2-12.