Method, device and storage medium for obtaining a high-precision estimate of yaw rate

Abstract

The present invention relates to a method for determining high-precision yaw rate information for controlling a vehicle, the method comprising the following steps: determining a first yaw rate estimate of the vehicle by fusing sensor data from an inertial sensor, a GNSS sensor, a wheel speed sensor, and / or a steering angle sensor; determining a second yaw rate estimate of the vehicle by evaluating sensor data from a camera associated with the vehicle, the camera optically sensing the vehicle's surroundings; correcting the first yaw rate estimate by means of the second yaw rate estimate to determine a corrected yaw rate estimate; and outputting the corrected yaw rate estimate as yaw rate information to generate a control signal for controlling the vehicle. The present invention also relates to a corresponding control device, a computer program, and a computer-readable storage medium.

Description

Technical Field

[0001] The present invention relates to a method and a control device for determining high-precision yaw rate information for controlling a vehicle. The present invention also relates to a computer program for implementing the aforementioned method and a computer-readable storage medium having a corresponding computer program. Background Art

[0002] Information about the vehicle's current yaw rate is required in various automotive applications. For example, during high-precision parking maneuvers or, more generally, in the context of automated driving, the yaw rate is a crucial parameter in various control systems used to guide the vehicle. Essentially, information about the vehicle's yaw rate must be available for any driving situation that may occur with the vehicle in question. In particular, automated driving requires extremely high yaw rate accuracy at all times. If the vehicle is in autonomous mode, characterized by the absence of any human driver input, the estimated yaw rate must always meet the required accuracy. This is because, in autonomous mode, the possibility of an emergency stop of the vehicle must always be present, as is the case, for example, if a relevant controller fails.

[0003] The vehicle's yaw rate can be measured directly using a suitable inertial measurement unit (IMU). However, such inertial sensors have high unreliability under various conditions, meaning that the individual measured values ​​do not always meet the high demands of autonomous driving. Due to these uncertainties, multiple information sources are used to improve signal quality. Existing methods for yaw rate estimation primarily use wheel speeds, accelerations, rotation rates, GNSS data, and GNSS correction data for individual wheels. These variables are correlated using specific methods and, for example, fused using specific algorithms to estimate the yaw rate as accurately as possible in various driving situations. The challenge here is ensuring the accuracy of the yaw rate estimate even when the sensor data is not of sufficient quality. Consequently, variables measured using inertial sensor systems, such as acceleration or rotation rate, have high uncertainties, due, for example, to thermal drift of the relevant sensors. Furthermore, the position data determined using GNSS sensors and the associated GNSS correction data may not be of sufficient quality due to reception difficulties. This is often the case, especially in urban areas, where many tall buildings cause reflections and, therefore, poor reception of GNSS and correction data signals. If extreme physical effects, such as high temperature increases, also occur in this situation, inertial sensors can also be highly unreliable. Summary of the Invention

[0004] The object of the present invention is therefore to provide a method for correcting a yaw rate estimate, in particular one calculated by fusing data from an inertial sensor system and GNSS, under conditions of high uncertainty in the corresponding sensors. This object is achieved by the method of the present invention. Furthermore, this object is achieved by the controller, computer program, and computer-readable storage medium of the present invention. Further advantageous embodiments are described in the preferred embodiment.

[0005] According to the present invention, a method for determining high-precision yaw rate information for controlling a vehicle is provided. A first yaw rate estimate of the vehicle is determined using sensor data fusion from a GNSS sensor, an inertial sensor, a wheel speed sensor, and / or a steering angle sensor. In a further method step, a second yaw rate estimate of the vehicle is determined by evaluating sensor data from a camera associated with the vehicle, which optically detects the vehicle's surroundings. In a subsequent method step, the first yaw rate estimate is corrected using the second yaw rate estimate to determine a corrected yaw rate estimate. Finally, the corrected yaw rate estimate is output as yaw rate information to generate control signals for controlling the vehicle. High-precision yaw rate information is achieved by using the second yaw rate estimate determined using optical observation of the surroundings to correct the first yaw rate estimate determined using sensor fusion. The yaw rate information thus obtained ensures high accuracy of the relevant control signals, particularly those required for autonomous driving modules. Furthermore, even in the event of failure of a single sensor, the method used ensures sufficiently high accuracy of the yaw rate to enable emergency measures, such as emergency stopping of the vehicle, to be taken.

[0006] In one specific embodiment, a first yaw rate estimate is compared with a second yaw rate estimate in order to determine a yaw rate correction value. The yaw rate correction value is then used to correct the first yaw rate estimate. The yaw rate correction value determined in this manner is a variable that essentially reflects the current error in the first yaw rate estimate. The temporal evolution of this error allows conclusions to be drawn as to whether a constant deviation or a deviation caused, for example, by thermal drift of the sensor signal is involved. This information can, in turn, be used to determine a corrected yaw rate estimate even if the second yaw rate estimate is temporarily unavailable.

[0007] In another specific embodiment, a deviation of the first yaw rate estimated value from the second yaw rate estimated value is determined as a yaw rate correction value. The yaw rate correction value is calculated using the first yaw rate estimated value in order to determine a corrected yaw rate estimated value. This approach allows for a particularly simple correction of the yaw rate estimated value.

[0008] In another specific embodiment, a second yaw rate estimate is determined with a time delay relative to the first yaw rate estimate. The first yaw rate estimate is temporarily stored. Furthermore, to determine the yaw rate correction value, the temporarily stored first yaw rate estimate is read and compared with the currently determined second yaw rate estimate. By temporarily storing the first yaw rate estimate, the time offset between the two yaw rate estimates can be compensated in a particularly simple manner. The yaw rate correction value determined in this manner also exhibits a time offset relative to the currently determined first yaw rate estimate. However, this is not a problem in principle given a constant error in the first yaw rate estimate and can be calculated, if necessary, even if the error undergoes changes due to thermal drift.

[0009] According to another embodiment, the second yaw rate estimate is used to verify the plausibility of the first yaw rate estimate. This makes it possible to monitor the sensors used for sensor fusion and the corresponding sensor fusion device in a particularly advantageous manner. This ensures the flawless function of the involved components.

[0010] In another specific embodiment, the last determined yaw rate correction value is stored. If no current yaw rate correction value is available, the last determined yaw rate correction value is read out again and used to correct the first yaw rate estimate. This allows the first yaw rate estimate to be corrected accordingly even if corresponding components, such as the correction value determination device, the corresponding device for determining the second yaw rate estimate, or the camera system, fail.

[0011] Furthermore, according to another specific embodiment, if the sensor data from the camera is not available or is not available with sufficiently high quality, the first yaw rate estimate is output as yaw rate information in order to generate a control signal for controlling the vehicle. In this way, even in the event of a camera failure or a failure of a component that determines the second yaw rate estimate from the camera sensor data, still usable yaw rate information can be provided.

[0012] According to another aspect, a control device for determining high-precision yaw rate information for controlling a vehicle is provided, wherein the control device is configured to implement at least part of the steps of the above method.

[0013] According to another aspect, a computer program is provided which comprises instructions which, when the computer program is executed on a computer, arrange for the computer to carry out the above-mentioned method.

[0014] Finally, according to another aspect, a computer-readable storage medium is provided, on which the above-described computer program is stored. Implementing the above-described method in the form of a computer program offers a particularly high degree of flexibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The present invention is explained in more detail below with reference to the accompanying drawings.

[0016] Figure 1 A vehicle is schematically shown, which has a sensor for ascertaining a certain variable and a control device for ascertaining an estimated value for the vehicle's yaw rate using the ascertained variable;

[0017] Figure 2 A block diagram shows a control device of a vehicle for determining a corrected yaw rate estimate using a camera-based yaw rate estimate; and

[0018] Figure 3 The flowchart of a method for correcting a yaw rate estimate ascertained by sensor mixing with the aid of a yaw rate estimate ascertained by optical sensing of the surroundings is schematically shown. DETAILED DESCRIPTION

[0019] Attachment Figure 1 A vehicle 100 is schematically shown, located on a road 210 in an environment 200. Vehicle 100, which is preferably designed as an automated vehicle, has a sensor system 130 comprising a plurality of different sensors 131, 133, 135, 137, 139, 141 for sensing certain vehicle parameters and its environment 200. In particular, in the present case, sensor system 130 comprises an inertial sensor system having at least one inertial sensor 131 for sensing acceleration and / or rotational movement of vehicle 100, a wheel speed sensor system having preferably a plurality of wheel speed sensors 133 for sensing the rotational speeds of individual wheels 101, 102, and a steering angle sensor for sensing the current steering angle. Vehicle 100 also has a GNSS sensor 137 for determining position within the context of satellite-assisted navigation, and a correction data receiver 139 for receiving corresponding GNSS correction data 140. Sensor device 130 also has at least one camera for optically sensing vehicle surroundings 200. In the present example, camera 141, designed as a front camera, has a forward-facing sensing area 144 and thus senses a section of road 210 located in front of vehicle 100 with corresponding road markings 211, 212, 213, as well as an object 220 located in front of vehicle 100.

[0020] As attached Figure 1 As can be seen, vehicle 100 also includes a control device 110 for controlling certain functions. Control device 110 is designed to evaluate data 132, 134, 136, 138, 140, 142 provided by various components 131, 133, 135, 137, 139, 141 and, based on this evaluation, to provide yaw rate information for controlling vehicle 100. In particular, control device 110 is designed to output a suitable signal 305 to at least one actuator control device 160, which causes the actuator control device to generate a corresponding control signal 306 for actuating certain actuators 161, 162 of vehicle 100.

[0021] As in Figure 1 As shown by way of example in FIG, the storage device 110 also includes a computer-readable storage medium 118 on which a computer program for executing the method described here can be stored.

[0022] Vehicle 100 is preferably configured as an automated vehicle 100, which can typically perform at least a portion of its driving tasks autonomously. Depending on the level of autonomy, automated vehicle 100 can either provide only assistance functions to the driver (SAE Levels 1 and 2) or automatically perform longitudinal and lateral guidance without any human intervention (SAE Levels 3 to 5). However, particularly in vehicles with SAE Levels 4 and 5 of autonomy without a driver, a very robust sensor system is crucial in order to reliably navigate the vehicle within its environment.

[0023] For this reason, important parameters of vehicle 100 are determined by jointly evaluating sensor data from multiple sensors. This ensures that the parameters determined based on the sensor data have the necessary accuracy. While it is generally possible to directly measure vehicle yaw rate 300, that is, the rotation of vehicle 100 about its vertical axis 103, using a suitable inertial measurement unit (IMU), the corresponding inertial sensors 131 have high unreliability under various conditions, so that individual measured values ​​do not always meet the high demands of autonomous driving. Due to these unreliability factors, multiple information sources are used in vehicle 100 to improve the quality of yaw rate information. Because different sensors, such as inertial sensor systems, GNSS, wheel speed sensors, steering angle sensors, etc., have partially complementary characteristics, it is useful to use multiple of these sensors simultaneously to determine yaw rate 300. An estimated value for yaw rate 300 is typically determined by fusing inertial sensor 131 with other sensors, in particular GNSS sensor 137 for satellite-assisted navigation, wheel speed sensor 133, and / or steering angle sensor 135. A corresponding fusion algorithm, such as a Kalman filter suitably implemented in control device 110, is used for sensor fusion. The more accurately yaw rate 300 can be determined, the higher the performance achieved by the system and algorithm specified therein.

[0024] However, the sensor data 132, 134, 136, 138, and 140 used for sensor fusion are subject to varying degrees of unreliability due to typical effects specific to the respective sensor types. Thus, the reception of GNSS signals by the GNSS sensor 141 and the reception of correction data signals by the correction data receiver 150 may be disrupted by terrain conditions. Furthermore, the sensor data 134 from the inertial sensor 133 may also be highly unreliable due to temperature variations.

[0025] According to the present invention, a camera-based yaw rate estimate is therefore considered to improve the accuracy of the sensor-based yaw rate estimate. Since the camera-based yaw rate estimate is less sensitive to temperature, it is a more reliable source of information for deriving the yaw rate.

[0026] Figure 2 Schematically shown Figure 1Detailed view of control device 110 of vehicle 100 in FIG. Control device 110 includes a yaw rate determination device 111 for determining high-precision yaw rate information, a calculation device 150 for determining a camera-based yaw rate estimate 302 using sensor data 142 from camera 140, a correction value determination device 115 for determining a yaw rate correction value 303, and optionally a buffer 117 for temporarily storing yaw rate correction value 303. Yaw rate determination device 111 includes a sensor fusion device 112 for determining a first yaw rate estimate 301, a correction device 114 for determining a corrected yaw rate estimate 304, and a decision device 113 for outputting yaw rate information. Yaw rate determination device 111 first determines first yaw rate estimate 301 from sensor data 132, 134, 136, 138 received from sensors 131, 133, 135, 137 using a suitable fusion algorithm. Next, yaw rate determination device 111 outputs the determined first yaw rate estimate 301 to correction device 114, decision device 113, and correction value determination device 115. Parallel to the determination of first yaw rate estimate 301, computing device 150 determines second yaw rate estimate 302 based on sensor data 142 from camera 140 and also outputs this second yaw rate estimate to correction value determination device 115. Correction value determination device 115 then determines a possible deviation of first yaw rate estimate 301 from the actual value of yaw rate 300 based on first yaw rate estimate 301 and second yaw rate estimate 302, and then generates yaw rate correction value 303 corresponding to the determined deviation. This yaw rate correction value is then output to correction device 114 in a manner similar to first yaw rate estimate 301. Since second yaw rate estimate 302 has a certain time delay relative to first yaw rate estimate 301 determined by sensor fusion due to the higher computational complexity in evaluating sensor data 142 provided by camera 41, first yaw rate estimate 301 may be delayed by a corresponding time period before being compared with second yaw rate estimate 302 in correction value determination device 115. This can be done, for example, by temporarily storing first yaw rate estimate 301 in a suitable storage device 116 and reading it out again from this storage device after a defined time has elapsed. Figure 2 In the embodiment, such a storage device 116 is arranged in correction value determination device 115 . In principle, however, the storage device can also be provided at another location in control device 110 , for example in yaw rate determination device 111 .

[0027] Once first yaw rate estimate 301 and correction value 303 are present in correction device 114, the correction device corrects first yaw rate estimate 301 and outputs the corresponding correction result to decision device 113 in the form of corrected yaw rate estimate 104. Here, decision device 113 determines, based on a predetermined criterion, which of the two yaw rate correction values ​​301 and 304 should be output as yaw rate information 305. In particular, if a significant deviation is detected between first yaw rate estimate 301 and corrected yaw rate estimate 304, decision device 113 may output corrected yaw rate estimate 304 as yaw rate information 305. Conversely, if no deviation or only a very small deviation is detected between first yaw rate estimate 301 and corrected yaw rate estimate 304, decision device 113 may output first yaw rate estimate 301, which is slightly updated compared to corrected yaw rate estimate 304, as yaw rate information 305.

[0028] As from Figure 2 As can be seen in , yaw rate information 305 determined in this way can be transmitted to at least one actuator control unit 160 , which generates control signals 306 for actuating actuators 161 of vehicle 100 based on this information.

[0029] Optionally, the control device 110 may also include a buffer memory 117 for buffering the yaw rate correction value 303. Figure 2 As shown by way of example in , the buffer can be arranged between correction value determination device 115 and yaw rate determination device 111 or can also be designed as part of one of these components.

[0030] If current yaw rate correction value 303 is temporarily unavailable, yaw rate correction value 303 stored in intermediate memory 117 can be read again from intermediate memory 117 and used to correct first yaw rate estimate 101 by correction device 114 .

[0031] The control device 110 is typically implemented in the form of hardware. Figure 2 Depending on the application, at least some of the described functions can be implemented in hardware, software, or a hybrid of hardware and software. Implementing the corresponding functions in hardware offers a particularly efficient method of operation. Conversely, implementing the corresponding functions in software offers a particularly high degree of flexibility.

[0032] exist Figure 3A simplified flowchart 400 illustrating the method is shown in FIG. In method step 410 , a camera-based estimate of the vehicle's yaw rate is determined. This is accomplished by sensing objects and / or structures in surroundings 400 of vehicle 100 using at least one vehicle-mounted camera 130 and subsequently evaluating the data obtained. In method step 420 , a sensor-based estimate of the yaw rate of the vehicle in question is determined based on sensor fusion. To this end, sensor data is first sensed using various sensors of vehicle 100 , and then a sensor-based estimate of the yaw rate is determined by fusing the sensed sensor data. In method step 430 , a correction value 303 for sensor-based yaw rate estimate 301 is determined based on sensor-based yaw rate estimate 301 and camera-based yaw rate estimate 302 . This is accomplished, for example, by comparing sensor-based yaw rate estimate 301 with camera-based yaw rate estimate 302 . In a further method step 440, a corrected yaw rate estimate 304 is determined. This is typically done by determining sensor-based yaw rate estimate 301 using yaw rate correction 303. In a further method step 450, a decision is made as to whether corrected yaw rate estimate 304 or sensor-based yaw rate estimate 301 is to be used as yaw rate information. Subsequently, in a further method step 460, the respectively selected yaw rate estimate 301, 304 is output in the form of yaw rate information. Finally, in a further method step 470, at least one control signal 306 for controlling vehicle 100 is generated based on the output yaw rate information.

[0033] The approach described herein uses a camera-based yaw rate estimate to improve the accuracy of a sensor-based yaw rate estimate. Because the camera-based yaw rate estimate is less sensitive to temperature, it is a more reliable source of information for determining the yaw rate. However, due to computationally intensive camera data processing, the camera-based yaw rate estimate has a relatively high time delay. Consequently, the camera-based yaw rate estimate cannot be directly used for fusion with other yaw rate sources, but is well-suited for correcting sensor uncertainties underlying the sensor-based yaw rate estimate 301. If, for example, the yaw rate estimate from the fusion algorithm exhibits a high offset, this can be detected by comparison with the camera-based yaw rate estimate in a correction algorithm. If the high quality of the camera information is ensured during this time, a correction of the sensor-fusion-based yaw rate estimate can be performed. If, for example, a correction value from the correction algorithm is available, it can be used, particularly during periods of GNSS signal unavailability, when camera information is missing (e.g., when lane markings are absent or unrecognized), or even in the event of a failure of one of these components. In a further application, instead of using a corrected yaw rate estimate, a sensor-based yaw rate estimate is used, which was calculated beforehand with the aid of a yaw rate correction value.

[0034] In addition to improved estimation accuracy, ensuring yaw rate accuracy during camera failures and / or GNSS reception disturbances provides another advantage. This is particularly important when the failure or disturbance persists over a long road section. This can be the case during an autonomous emergency stop. This function is activated, for example, when one or more hardware faults occur. In such scenarios, only a few sensors are available for determining the yaw rate, making the correction value extremely important.

[0035] The core of the approach described here is to use camera information to correct for the unreliability of highly accurate yaw rate estimates. This unreliability can arise from the varying sensitivities of different sensor data. Therefore, the accuracy of inertial data is typically temperature-dependent. In contrast, GNSS sensor data is often highly inaccurate in cities, as GNSS signal reception is disrupted by reflections from buildings. In the near future, autonomous vehicles will also operate in cities and in extreme temperatures, necessitating the use of additional sensors for fusion or correction. The fusion of sensor data from the inertial sensor system with sensor data from the GNSS sensor and correction data serves as the basis for the yaw rate estimate. In situations where the estimated yaw rate exhibits high uncertainty, the camera information is used to verify its plausibility and correct it.

[0036] Although the present invention has been described and illustrated in detail through preferred embodiments, the present invention is not limited to the disclosed embodiments. In addition, other variations can be derived by those skilled in the art without departing from the scope of protection of the present invention.

Claims

1. A method for obtaining high-precision yaw rate information to control a vehicle (100), the method comprising the following steps: - determining a first yaw rate estimate (301) of the vehicle (100) by means of a fusion of sensor data (132, 134, 136, 138) from an inertial sensor (131), a GNSS sensor (137), a wheel speed sensor (133) and / or a steering angle sensor (135), - determining a second yaw rate estimate (302) of the vehicle (100) by evaluating sensor data (142) of a camera (141) assigned to the vehicle (100), the camera optically sensing the surroundings (200) of the vehicle (100), - performing a correction of the first yaw rate estimate (301) by means of the second yaw rate estimate (302) in order to determine a corrected yaw rate estimate (304), and - outputting the modified yaw rate estimate (304) as yaw rate information in order to generate a control signal (306) for controlling the vehicle (100), wherein the first yaw rate estimate (301) is compared with the second yaw rate estimate (302) to obtain a yaw rate correction value (303), wherein the yaw rate correction value (303) is used to correct the first yaw rate estimate (301), The second yaw rate estimate (302) is determined with a time delay relative to a first yaw rate estimate (301), the first yaw rate estimate (301) being temporarily stored, and the temporarily stored first yaw rate estimate (301) being read out and compared with the currently determined second yaw rate estimate (302) in order to determine the yaw rate correction value (303).

2. The method according to claim 1, wherein A deviation of the first yaw rate estimate (301) relative to the second yaw rate estimate (302) is determined as a yaw rate correction value (303), wherein the yaw rate correction value (303) is calculated using the first yaw rate estimate (301) in order to determine a corrected yaw rate estimate (304).

3. The method according to claim 1 or 2, wherein: The second yaw rate estimation value (302) is used to verify the credibility of the first yaw rate estimation value (301).

4. The method according to claim 1 or 2, wherein: The last ascertained yaw rate correction value (303) is stored, wherein if no current yaw rate correction value (303) is available, the last ascertained yaw rate correction value (303) is read out and used to correct the first yaw rate estimate (301).

5. The method according to claim 1 or 2, wherein: The availability and quality of the sensor data (142) of the camera (141) are monitored, wherein the yaw rate correction value (303) is determined and / or used to correct the first yaw rate estimate (301) only when the sensor data (142) of the camera (141) are available and of sufficiently high quality.

6. The method according to claim 5, wherein: In the event that sensor data (142) of the camera (141) are not available or are not available with sufficiently high quality, the first yaw rate estimate (301) is output as yaw rate information in order to generate a control signal (306) for controlling the vehicle (100).

7. A control device (110) for determining high-precision yaw rate information for controlling a vehicle (100), the control device being configured to carry out at least part of the steps of the method according to one of claims 1 to 6.

8. A computer program product comprising a computer program, the computer program comprising instructions which, when the computer program is executed by a computer, arrange for the computer to carry out the method according to one of claims 1 to 6.

9. A computer-readable storage medium (118) on which a computer program is stored, the computer program comprising instructions which, when the computer program is executed by a computer, cause the computer to carry out the method according to one of claims 1 to 6.

Images