WO2023098958A1

WO2023098958A1 - Method for a precise odometry in the event of a brake slip

Info

Publication number: WO2023098958A1
Application number: PCT/DE2022/200267
Authority: WO
Inventors: Kai Bretzigheimer; Mohanad Youssef
Original assignee: Continental Automotive Technologies GmbH
Priority date: 2021-12-02
Filing date: 2022-11-14
Publication date: 2023-06-08
Also published as: DE102021213696A1

Abstract

The invention relates to a method for determining the position of a vehicle, in particular in order to correct a vehicle speed and in order to correct the wheel speeds. A simple method which can function with little sensor data is to be provided in order to correct a slip of a vehicle. According to the invention, the method has the following steps: a) recording speed signals of the wheels and calculating the changes in the speed, b) carrying out a low-pass filtering process on the speed signals, c) limiting the speed signals to speed signals with a change in speed below an acceleration threshold A2, d) calculating the vehicle speed change from the change in speed of the fastest speed signal of the limited and filtered speed signals, e) calculating the wheel slip from the difference between the unfiltered speed signal and the fastest limited and filtered speed signal, f) determining a slip distance if the value of the difference lies above a threshold, and g) calculating the actual distance which has been traversed by the vehicle by adding the distance traversed by all of the non-slipping wheels, adding a specified fraction of the calculated slip distance for each non-slipping wheel, and dividing the sum by the number of non-slipping wheels.

Description

Procedure for accurate odometry in brake slip

The present invention relates to a method for determining the position of a vehicle, in particular for correcting a vehicle speed and for correcting the wheel speeds.

DE 199 36 710 A1 discloses a method for determining the vehicle speed using a vehicle dynamics system (ABS, anti-lock braking system). This document aims to determine wheel slip under hard braking at the level of road grip.

There are several studies and methods in the literature for determining the wheel slip of a moving vehicle on the road. In most of these methods, however, other sensors are used in addition to the wheel speed sensors, which make the system even more complex, such as B. the Inertial Measurement Unit (IMU) and/or positioning sensors (e.g. GPS, Global Positioning Satellite) to use the acceleration data to determine how much the vehicle has actually moved and to calculate the wheel revolutions measured by the wheel speed sensors compare. Relevant basics are z. B. Described in Chris C. Ward and Karl Lagnemma, "Model-Based Wheel Slip Detection for Outdoor Mobile Robots", 2007 IEEE International Conference on Robotics and Automation.

The problem with the approach presented in DE 199 36 710 A1 is that the torque induced at the wheels during the braking process must be known. This signal can be determined in a brake system. However, it is usually not available in a driver assistance system (also Engi. ADAS=Advanced Driver Assistance System). In order to determine the braking torque from the brake pressure, information about the dimension of the brakes is required, which is only available in the braking system. In addition, this algorithm determines the vehicle speed, while more importantly, the to know the actual distance traveled by the vehicle. This approach therefore requires greater networking of the vehicle systems and additional sensor information and is therefore relatively complex and not suitable for every combination of driver assistance system and braking system.

In autonomous driving and in particular in automated parking, vehicle odometry is a critical and essential component to locate the vehicle in the environment, and many other components depend on its accuracy (e.g. environmental modelling, computer vision, traction control, etc .). The job of the odometry component is to use the onboard sensors (wheel speed sensors, inertial sensors, GPS, etc.) to estimate the amount of vehicle movement and rotation with the accuracy required by the user components. One of the requirements of odometry is to accurately determine the movement and rotation of the vehicle even under different environmental conditions (e.g. snow, icy roads, etc.). At the same time, it is desirable, albeit with a limited amount of sensor information (e.g. for vehicles with a basic equipment, i.e. fewer sensors) and / or in the event of failure of one or more of the additional sensors (inertial measuring devices, GPS, braking data from the braking system, etc.), a satisfactory one Position determination by the driver assistance system is possible.

A problem for reliable attitude determination from wheel speed sensors alone is the detection of wheel slip while the vehicle is slowed down either by hard braking or on low-friction roads at the limit of road grip. This phenomenon degrades the accuracy of the odometry as the wheel rotation is not realistic compared to the actual vehicle motion. As a result, one or more wheels lose traction temporarily and may move faster (under low friction) or slower (under heavy braking) than would be expected based on the vehicle's actual distance travelled. The object of the invention is to provide a method for determining and correcting wheel slip that is as simple as possible.

According to the invention, a method of the type mentioned is provided, comprising the following steps: a) recording speed signals from the speed sensors of several wheels of the vehicle and calculating the speed changes for each speed signal, b) applying low-pass filtering to the individual speed signals of the wheels to eliminate noise in the signals to filter out speed changes above a first acceleration threshold Ai and to obtain filtered speed signals, c) limiting to those of the filtered speed signals whose speed change is below a second acceleration threshold A2, d) calculating the vehicle speed change from the speed change of the fastest of the limited and filtered speed signals, e ) Calculating the wheel slip of each wheel by forming the difference between the respective unfiltered, unlimited speed signal and the speed signal of the fastest of the limited, filtered speed signals, f) determining for each of the wheels that the wheel is slipping if the amount of the respective difference formed in e) is above a limit, and if no, calculating a slip distance traveled for each non-slipping wheel, g) calculating the actual distance traveled by the vehicle by:

- summing the distance traveled of all wheels not determined to be slipping, and

- adding a predetermined fraction of the calculated slip distances for each wheel not determined to be slipping, and

- Divide this sum by the number of non-slipping ones

Wheels. The method according to the invention aims to improve the accuracy of the vehicle odometry in a simple manner (that is to say with fewer sensors), particularly in driving situations in which the vehicle is braked at the limit of road adhesion or road adhesion is reduced for other reasons. This can be the case either with heavy braking or with low friction of the road surface (e.g. due to leaves, water, ice, etc.). The wheels can be braked into slip, meaning they rotate slower than would be expected at vehicle speed. At the same time, they are individually modulated by the anti-lock braking system (ABS). Taking into account the wheel speed of the wheels with strong slip when determining the distance covered leads to a deterioration in the accuracy of the vehicle odometry.

The solution according to the invention now proposes a simple method for identifying wheels with strong slip and for temporarily not including their associated route in the route calculation. At the same time, however, the other wheels also regularly exhibit slight slip, particularly in braking situations, which also falsifies the measured distance compared to the distance actually covered. Therefore, according to the invention, a predetermined fraction of the slip distances calculated for these wheels is added for those wheels that are not determined to be (strongly) slipping. The inventors have found that this procedure can be used to approximately determine the route actually covered with good accuracy. In contrast to the prior art, no data from the braking system or additional sensor data such as from inertial sensors or GPS are required. The method is therefore suitable both for the simplest and most cost-effective driver assistance systems possible and as redundancy for more complex systems, e.g. B. to bridge a sensor failure.

The method is primarily aimed at four-wheeled vehicles such as passenger cars, but can in principle also be used for vehicles with a different number of wheels. References to the front wheels / rear wheels are therefore to be understood as illustrative and not restrictive. The method is preferably carried out iteratively at fixed time intervals, e.g. B. every 5-20 ms, which can also correspond to the sampling rate of the wheel speed sensors. At least the previously measured speed value of the respective wheel is used for the calculation of speed changes/accelerations.

First, in step b), the wheel speed signals are filtered in order to reduce the measurement noise. The time constant of the filter is selected in such a way that unphysically high changes in speed are filtered out, but changes in speed that could be associated with wheel slip are retained. As a result, the method can be made considerably more reliable.

In the next step c), the filtered speed signals whose speed change is below a second acceleration threshold A2 are limited (limited). This occurs because a typical road vehicle cannot decelerate or accelerate more than a certain amount of acceleration. If the wheel speed is changing more rapidly, this must be due to wheel slip (where the filtering has already removed potential signal noise). In the following, limited speed signals are therefore speed signals that are below the second acceleration threshold.

In step d), the fastest wheel is searched for among the filtered and limited speed signals. This is the wheel with the least amount of slip and is the best choice for determining actual vehicle speed.

In e) it is calculated how large the wheel slip of each wheel is by forming the difference between the respective unfiltered, unlimited speed signal and the speed signal of the fastest of the limited, filtered speed signals. Finally, in step f) it is determined for each of the wheels that the wheel is slipping if the amount of the respective difference formed in e) is above a limit value, and if not, a slip distance covered is calculated for each non-slipping wheel. This ensures that no signal noise or small deviations in the wheel speeds are identified as relevant slip, but only slip of a relevant magnitude. As described above, the wheels determined to be non-slip also regularly exhibit slight slip (especially in braking situations).

Finally, using the data determined in the previous steps, in step g) the actual distance traveled by the vehicle is calculated by adding the distance traveled for all wheels not determined to be slipping and adding a predetermined fraction of the calculated slip distances for each wheel not determined to be slipping wheel, and dividing this sum by the number of non-slipping wheels. The inventors have found that the actual distance traveled by the vehicle can be approximated fairly accurately by calculating a corrected average of the distance traveled for all non-slipping wheels which are (at least in some situations, more on that later) a fraction of the calculated slipping distances of the non-slipping wheels is included. The heavily slipping wheels are therefore not included in the calculation and at the same time a correction for the slight slippage of the other wheels (not determined as slipping) is calculated.

It is preferred if in step g) the specified fraction of the calculated slip distances is between 4% and 20%, preferably between 7% and 15%, particularly preferably about 8%. If the vehicle is braked but none of the wheels are slipping, this is an indication that the grip level has not been reached. However, hard braking slip can still occur, which is still below the grip level. This slippage can already affect the accuracy of vehicle odometry since the wheels are already turning slower than the vehicle is moving. Here "high friction" is assumed and a linear characteristic of the corresponding p-slip curve is used. The p-slip curve of the current road friction cannot be selected because in general neither the road surface nor the braking force is known. However, for simplification, a linear characteristic can be assumed and slip becomes a linear function of deceleration. Roughly approximated, the slip at 10 m/s ² is about 8% and at 5 m/s ² 4% with these simplifying assumptions. If one of the wheels is slipping, the grip level limit should be reached. If 10 m/s ² is assumed to be the maximum practical braking acceleration, this results in a predetermined fraction of about 8% for the slip correction. This slip is added to the individual wheel displacements of the non-slipping wheels. However, an improvement in the odometry can also be achieved in a broader range of parameters.

In the context of this application, "about" z. B. be understood that a deviation of ±10% from the stated value should still be included.

Preferably, after step c), the following steps are performed: c1) determining that a wheel is braked if the speed of a filtered and limited speed signal decreases, and c2) determining that the vehicle is braked if more in step c1). when two wheels were determined to be braked. If the wheel speed becomes progressively slower in the subsequent journals of the algorithm, the wheel is considered to be braked. If at least two wheels slow down, the entire vehicle is deemed to have braked. In the following, this is important in order to recognize and differentiate between wheel spin when accelerating and brake slip.

Alternatively, the determination that the vehicle is being braked is made by information obtained from the brake system. The information can be obtained directly or indirectly from the braking system. For example, the activation of the vehicle's brake light switch or the information about a measured brake pressure above a limit value, which is generated by the brake pedal or a braking assistance function, can be evaluated. In one embodiment, steps e), f) and g) will only be performed if it has been determined that the vehicle is being braked, otherwise in a h) calculating the actual distance traveled by the vehicle by: - adding up the distance traveled of all limited wheels, and

- Divide this sum by the number of all limited bikes.

If a braking situation is detected, the method according to the invention is carried out as described in steps e-g). However, if no clear braking situation is determined (e.g. slip when accelerating or slip of the wheels in general due to poor road grip), it has been found that it is most advantageous to only calculate those wheels whose speed change is above the second acceleration threshold calculated from the distance actually traveled by the vehicle and not to undertake any additional slip correction as in step g).

After step a), normalized speed signals of the wheels are preferably used throughout the method, which are calculated as follows from the measured speed signals of the wheels: a1) recording a rear axle speed signal in the middle of a vehicle rear axle and at least one steering angle of the front wheels, a2) calculating a normalized speed signal for each of the wheels from the measured speed signals and the rear axle speed signal and at least one steering angle of the front wheels, so that differences in the distance traveled by the wheels due to cornering of the vehicle are reduced.

While the vehicle is cornering, the wheel speeds are different because the wheels are turning on different radii of the curve. The wheel on the outside of the curve spins faster than the wheel on the inside of the curve. If this is not corrected, wheel slip could be incorrectly detected. For this reason, the wheel speed of the individual wheels is normalized to the corresponding speed at the center of the rear axle. Such corrections to the wheel speeds in curves are known in principle to those skilled in the art and can be carried out in various ways (with different accuracy) are carried out. This is explained in more detail later with reference to FIG. 2 . However, this embodiment synergistically improves the accuracy of the method according to the invention, since the detection of slip is much more reliable if this source of wheel speed difference is eliminated (or reduced).

In one embodiment, the low-pass filtering uses a PT1 filter, which preferably has a time constant T of 0.05 s<T<0.01 s, particularly preferably about T=0.02 s. A PT1 filter (also called a PT1 element) is known in control engineering as a simple low-pass filter and is ^completely sufficient for the relatively coarse noise filtering required here for filtering out extremely fast changes in speed, which can practically only be attributed to noise.

The first acceleration threshold Ai is preferably in the range of 50 m/s ² <Ai <200 m/s ² , particularly preferably around 100 m/s ² . The fact that speed changes above such a first acceleration threshold Ai are filtered out is to be understood here and also more generally in such a way that, for example, such high-frequency signal components are suppressed by at least 10 db, preferably by at least 20 db, particularly preferably by at least 30 db, by the low-pass filtering for frequency components , which correspond to changes in speed above the first acceleration threshold Ai.

It is preferred if the second acceleration threshold A2 is in the range of 8 m/s ² <A2 <12 m/s ² , particularly preferably around 10 m/s ² . Accelerations that a car can carry out with full road grip usually end in this range, and higher accelerations of a wheel are therefore usually due to slippage of the corresponding wheel.

The invention also relates to a driver assistance system that is set up to carry out a method according to one of the above embodiments. The driver assistance system can use appropriate software or firmware for this purpose which, when executed, performs such a method. However, it is also conceivable to integrate an application-specific integrated circuit into the driver assistance system, which is set up to carry out the method according to the invention.

The characteristics, features and advantages of the present invention described above, as well as the manner in which they are achieved, will become clearer and more clearly understood in connection with the following description of the embodiments, which are explained in connection with the drawings.

Show it:

1 shows an exemplary flow chart of a method according to the invention,

2 shows a schematic representation of the wheel speed correction when cornering.

1 shows an embodiment of a method according to the invention. When a journal (call, iteration, etc.) of the method is started, a loop runs first for each wheel, starting with step 100 and ending with step 140, until the loop has run for all wheels.

In step 100, the speed signal of the associated speed sensor is recorded for the respective wheel.

In step 110, a rear axle speed signal is then recorded in the center of a vehicle rear axle and at least one steering angle of the front wheels, and a normalized speed signal for the wheel is calculated from the measured speed signal and the rear axle speed signal and at least one steering angle of the front wheels, so that differences in the traveled distance of the wheels can be reduced due to cornering of the vehicle. In step 120, low-pass filtering is applied to the single normalized wheel speed signal to filter out noise in the signal with speed changes above a first acceleration threshold Ai (e.g. 100 m/s ² ) and obtain a filtered (normalized) speed signal.

In step 130, the filtered speed signals are limited to those whose speed change is below a second acceleration threshold A2 (eg 10 m/s ² ).

In step 140 it is determined that a wheel is braked if the speed of the filtered and limited speed signal is decreasing. A comparison is made here with at least one previous speed value of the same wheel in order to calculate the change in speed for the corresponding wheel from the time profile of the speed signal. Wheels whose speed change is above A2 are therefore not determined as being braked.

When the 100-140 loop is completed for all wheels, in step 150 the fastest wheel is searched for among the normalized and filtered speed signals. In step 160, a limit is also set here to those wheels whose speed change is below A2. Then, in step 170, the vehicle acceleration (or deceleration) is calculated from the velocity change of the fastest of the limited, filtered, and normalized wheels, since this regularly best represents the actual acceleration of the vehicle.

In step 180 it is then determined that the vehicle as a whole is decelerating if more than two wheels were determined to be braking in step 140 of the respective loop.

From 190 to 240/250, a loop runs over all wheels, which differs primarily in whether it was determined in step 180 whether the vehicle brakes or not. If it was not determined that the vehicle is braking, then in step 200 for all limited wheels (speed change <A2) to calculate the distance actually driven by the vehicle, the distance of the respective wheel resulting from the filtered and normalized wheel speed is added and in step 260 by divides the number of these limited wheels to calculate an average actual distance traveled by the vehicle (excluding the wheels that are slipping).

If, on the other hand, it is determined in step 180 that the vehicle is braking, a correction is made in loop 190 to 240/250 in order to take account of the slip contributions to the distance traveled in the event of heavy braking. In step 210, the wheel slip is then determined for each wheel by forming the difference between the respective unfiltered (but normalized) speed signal and the speed signal of the fastest of the limited, filtered (and normalized) speed signals.

In step 220 it is then determined, for each of the wheels, that the wheel is slipping if the amount of the respective difference formed in step 210 is above a limit value.

In step 230, a distinction is then made for each wheel as to whether it was determined as slipping in step 220 and, if not, the entire wheel travel distance is calculated from the normalized and filtered wheel speed and added to calculate the actually traveled distance of the vehicle. Then, for each wheel not determined to be slipping, a predetermined fraction (e.g., 8%) of the slip distance calculated for that wheel (in step 210) is further added to calculate the vehicle's actual travel distance. The sum of wheel travel contributions thus obtained is then divided by the number of non-slipping wheels in step 260 to calculate an average actual distance traveled by the vehicle (with reduced inclusion of the slipping wheels).

In Fig. 2, a vehicle 1 with wheels 2, 3, 4, 5 is shown schematically from above to show a possible way of wheel speed correction when cornering as in Step 110 described to explain. As can be seen, the wheels 2, 3, 4, 5 travel different distances when cornering and therefore also have different speeds, all of which deviate from the vehicle speed.

To correct the wheel speeds, a virtual front wheel angle a _c , ie the wheel angle of a virtual central wheel 6 , is determined from the steering wheel angle, ie the current steering wheel position. For this purpose, this determination uses a characteristic conversion curve, since the gear ratio depends on the steering wheel angle. In this case, the transmission ratio is greater for smaller steering wheel angles than for large steering wheel angles.

The wheel angles OFR, OFL of the two front wheels 2, 4 are now calculated from this virtual front wheel angle a _c together with the track width of the rear axle tw _re ar and the wheelbase W, using the formula

The plus sign applies to the front wheels and the minus sign to the rear wheels. The indices FL (front left), FR (front right), RL (rear left) and RR (rear right) each refer to one of the wheels 2, 3, 4, 5.

The curvatures K _FL , K _FR , K _RL , K _RR , K _C of the individual movement paths of the wheels 2, 3, 4, 5 and the virtual movement path of the central wheel 6 can now be calculated from these angles with: sin a _FL

^{KFF =} —W~ sin a _FR K ^{fr = -} iv “ tan a _FL

^Krl ~ W

Curve factors k _t can now be calculated from this, which result from:

J ^K RL -RL ~

^K c

The curve factors k _t indicate conversion factors with which the

Angular speeds of the individual wheels 2, 3, 4, 5 to a corresponding

Angular velocity of the virtual central wheel 6 is converted.

As a result, the differences in angular velocities, which are due to the

Cornering are calculated from the measured values.

The wheel speeds ^v whi,norm,i normalized to the reference point result ^from the measured wheel speeds v _wfll>i via: vwhl,norm,FL=k _FL -v _{whi FL} whl,norm,FR=k _FR -v _whi p _R vwhl,norm,RL=k _RL -v _whi:RL vwhl,norm,RR=k _RR -v _{whi RR}

Although the present invention has been illustrated and described in detail by preferred embodiments, it is not by those disclosed

examples limited. Reference List

100 step

110 step

120 step

130 step

140 step

150 step

160 step

170 step

180 step

190 step

200 step

210 step

220 step

230 step

240 step

250 step

260 step

Ai first acceleration threshold

A2 second acceleration threshold

T time constant

1 vehicle

2 wheels

3 wheels

4 wheels

5 wheels

6 virtual wheel

W wheelbase a _FL angle a _FR angle a _c Angle twrear track width of the rear axle

Claims

patent claims

1 . Method for determining the position of a vehicle (1), in particular for correcting a vehicle speed and for correcting the wheel speeds, comprising the following steps: a) recording speed signals from the speed sensors of a plurality of wheels (2, 3, 4, 5) of the vehicle (1) and calculating of the speed changes for each speed signal, b) applying low-pass filtering to the individual speed signals of the wheels (2, 3, 4, 5) in order to filter out noise in the signals with speed changes above a first acceleration threshold Ai and to obtain filtered speed signals, c) limiting to such of the filtered speed signals whose speed change is below a second acceleration threshold A2, d) calculating the vehicle speed change from the speed change of the fastest of the limited and filtered speed signals, e) calculating the wheel slip of each wheel (2, 3, 4, 5) by forming the difference between the respective unfiltered speed signal and the speed signal of the fastest of the limited, filtered speed signals, f) determining, for each of the wheels (2, 3, 4, 5), that the wheel (2, 3, 4, 5) is slipping if the amount of each difference formed in e) is above a limit value, and if not, calculating a slip distance covered for each non-slip wheel (2, 3, 4, 5), g) calculating the distance actually traveled by the vehicle (1) by:

- summing the distance traveled of all wheels (2, 3, 4, 5) not determined to be slipping, and

- adding a predetermined fraction of the calculated slip distances for each wheel (2, 3, 4, 5) not determined to be slipping, and - Divide this sum by the number of non-slipping wheels (2, 3, 4, 5). Method according to claim 1, wherein in step g) the predetermined fraction of the calculated slip distances is between 4% and 20%, preferably between 7% and 15%, particularly preferably about 8%. Method according to claim 1 or 2, wherein after step c) the following steps are performed: c1) determining that a wheel (2, 3, 4, 5) is braked if the speed of a filtered and limited speed signal decreases, and c2 ) Determine that the vehicle (1) is braked if in step c1) more than two wheels (2, 3, 4, 5) were determined as being braked. Method according to one of the preceding claims, wherein the determination that the vehicle (1) is being braked is made by information obtained from the braking system. Method according to one of the preceding claims, wherein steps e), f) and g) are only carried out if it has been determined that the vehicle (1) is braked, and otherwise in h) calculating the actually traveled distance of the vehicle ( 1 ) by: - Adding the distance traveled by all limited bikes, and

- Divide this sum by the number of all limited bikes. Method according to one of the preceding claims, wherein after step a) normalized speed signals of the wheels (2, 3, 4, 5) are used throughout the method, which are calculated as follows from the measured speed signals of the wheels (2, 3, 4, 5). : a1) recording a rear axle speed signal in the center of a vehicle rear axle and at least one steering angle (OFL, QFR) of the front wheels (2, 4), 18 a2) Calculation of a normalized speed signal for each of the wheels (2, 3, 4, 5) from the measured speed signals and the rear axle speed signal and at least one steering angle (OFL, OFR) of the front wheels (2, 4), so that differences in the Traveled distance of the wheels (2, 3, 4, 5) are reduced due to cornering of the vehicle (1). Method according to one of the preceding claims, wherein the low-pass filtering uses a PT1 filter which preferably has a time constant T with 0.05 s < T < 0.01 s, particularly preferably about T = 0.02 s. Method according to one of the preceding claims, wherein the first acceleration threshold Ai is in the range 50 m/s ² < Ai < 200 m/s ² , particularly preferably around 100 m/s ² . Method according to one of the preceding claims, wherein the second acceleration threshold A2 is in the range 8 m/s ² < A2 < 12 m/s ² , particularly preferably around 10 m/s ² . Driver assistance system that is set up to carry out a method according to any one of the preceding claims.