Vehicle motion control method and device, electronic equipment, vehicle and storage medium

Abstract

This invention provides a vehicle motion control method, device, electronic device, vehicle, and storage medium. The vehicle motion control method includes acquiring real-time vehicle driving data; inputting the real-time vehicle driving data into a motion control application approximation function to obtain a motion control quantity; inputting the motion control quantity into a performance evaluation application approximation function to obtain a motion controller performance evaluation index; optimizing the motion control application approximation function based on the motion controller performance evaluation index, and controlling the vehicle based on the motion control quantity output by the optimized motion control application approximation function; the performance evaluation application approximation function and the motion control application approximation function can be updated online, enabling the gradual learning of a motion control method suitable for the real vehicle operating environment using selected real driving data even without prior information from an accurate vehicle dynamics model, and ensuring the optimality of the method by updating the control method online.

Description

Technical Field

[0001] This invention relates to the field of intelligent driving technology, and in particular to a vehicle motion control method, device, electronic equipment, vehicle, and storage medium. Background Technology

[0002] Vehicle motion control includes vehicle path control and speed control. For articulated trucks, the design of their autonomous driving motion control algorithms needs to consider both the expected speed and the travel paths of the trailer and trailer. The goal is to minimize the deviation between the actual and expected travel paths of the trailer and trailer while maintaining the articulated truck's speed, achieving lateral and longitudinal motion control. Traditional autonomous driving motion control algorithms mainly include expert system control, PID control, linear quadratic programming, and model predictive control. These algorithms are designed based on expert experience rules or simplified vehicle models, and their performance is highly dependent on the completeness of the experience rules and the accuracy of the vehicle model control. Furthermore, because these algorithms are designed based on conventional operating conditions, their control performance is poor for special operating conditions such as "frequent braking" and "dragging" that occur during actual operation of articulated trucks, making it difficult to meet the road traffic safety requirements of autonomous articulated trucks during high-speed long-distance transportation. Summary of the Invention

[0003] This invention provides a vehicle motion control method, device, electronic device, vehicle, and storage medium to address the shortcomings of existing vehicle motion control methods, such as the inability to continuously optimize control accuracy, poor control performance under special working conditions, and difficulty in meeting the road traffic safety requirements of autonomous articulated trucks during high-speed long-distance transportation.

[0004] This invention provides a vehicle motion control method, comprising:

[0005] Obtain real-time vehicle driving data;

[0006] The real-time driving data of the vehicle is input into the approximate function of the motion control application to obtain the motion control quantity;

[0007] The motion control quantity is input into the performance evaluation application approximation function to obtain the motion controller performance evaluation index;

[0008] The motion control application approximation function is optimized based on the motion controller performance evaluation index, so as to control the vehicle based on the motion control quantity output by the optimized motion control application approximation function;

[0009] Wherein, the performance evaluation application approximation function is updated online based on the performance evaluation training approximation function with the same parameters, and / or, the motion control application approximation function is updated online based on the motion control training approximation function with the same parameters and the performance evaluation training approximation function.

[0010] According to a vehicle motion control method provided by the present invention, the performance evaluation application approximation function is updated online based on a performance evaluation training approximation function with the same parameters, specifically including:

[0011] Obtain training data for the performance evaluation training approximation function from the training database. The training data for the performance evaluation training approximation function includes training data for motion control quantities and their corresponding labeled values ​​for motion control evaluation indicators.

[0012] The motion control quantity training data is input into the performance evaluation training approximation function to obtain the predicted value of the motion controller performance evaluation index.

[0013] The loss function of the performance evaluation training approximation function is solved using the gradient descent method to update the parameters of the performance evaluation training approximation function. The loss function of the performance evaluation training approximation function is constructed based on the predicted value of the motion controller performance evaluation index and the labeled value of the motion control evaluation index.

[0014] The parameters of the updated performance evaluation training approximation function are reduced by a preset ratio and then assigned to the performance evaluation application approximation function to update the performance evaluation application approximation function online.

[0015] According to a vehicle motion control method provided by the present invention, the motion control quantity training data includes:

[0016] The vehicle's future trajectory data and acceleration data at the next moment.

[0017] According to a vehicle motion control method provided by the present invention, the method for obtaining the predicted value of the motion controller performance evaluation index includes:

[0018] The vehicle acceleration data is input into the performance evaluation training approximation function to obtain the first motion control evaluation index, which is the deviation between the vehicle speed and the target speed.

[0019] The vehicle's future trajectory data is input into the performance evaluation training approximation function to obtain a second motion control evaluation index. The second motion control evaluation index is the deviation between the vehicle's future trajectory output by the motion controller and the actual future trajectory of the vehicle during operation.

[0020] The first motion control evaluation index and the second motion control evaluation index are normalized and then summed. The cumulative value of the summation over a period of time is selected as the predicted value of the motion controller performance evaluation index.

[0021] According to a vehicle motion control method provided by the present invention, the motion control application approximation function is updated online based on a motion control training approximation function with the same parameters and a performance evaluation training approximation function, including:

[0022] The training driving data is obtained from the training database and input into the motion control training approximation function to obtain motion control quantity training data.

[0023] The performance evaluation training approximation function is used to calculate the gradient of the motion control quantity training data to obtain the performance evaluation gradient.

[0024] Multiply the performance evaluation gradient by the original motion control gradient of the motion control training approximation function to obtain the motion control gradient of the motion control training approximation function.

[0025] The loss function of the motion control training approximation function is solved based on the motion control gradient of the motion control training approximation function to update the parameters of the motion control training approximation function. The loss function of the motion control training approximation function is constructed based on the motion control quantity training data and the motion control quantity labeled data.

[0026] The parameters of the updated motion control training approximation function are reduced by a preset ratio and then assigned to the motion control application approximation function to update the motion control application approximation function online.

[0027] According to a vehicle motion control method provided by the present invention, the training driving data includes:

[0028] The historical trajectory data of the vehicle, the historical trajectory data of the surrounding vehicle, and the future trajectory data of the surrounding vehicle obtained from the collected data include at least one of the following: the vehicle's speed, the vehicle's current acceleration, the yaw rate, the vehicle's position information, and the surrounding vehicle's position information.

[0029] According to a vehicle motion control method provided by the present invention, the method for establishing the training database includes:

[0030] Collect vehicle driving data under different driving conditions, including different driving road segments, different driving speeds, and different traffic volumes.

[0031] The vehicle driving data is recombined to obtain at least one data combination; the data combination includes the vehicle's historical trajectory data combination, the vehicle's weekly historical trajectory data combination, the vehicle's future trajectory data combination, the vehicle's speed data combination, the target speed data combination, the acceleration data combination, and the vehicle's weekly future trajectory data combination.

[0032] Data from different data combinations with a unified timestamp are aligned to obtain training driving data and motion control quantity training data.

[0033] A training database is constructed based on the training driving data and its corresponding motion control quantity annotation data, as well as the motion control quantity training data and its corresponding motion control evaluation index annotation values.

[0034] According to a vehicle motion control method provided by the present invention, when both the performance evaluation application approximation function and the motion control application approximation function are updated online, the method includes:

[0035] The performance evaluation application uses an approximation function, and the motion control application uses an approximation function, which is iteratively updated alternately at preset intervals.

[0036] The present invention also provides a vehicle motion control device, comprising:

[0037] The first acquisition module is used to acquire real-time vehicle driving data;

[0038] The second acquisition module is used to input the real-time driving data of the vehicle into the motion control application approximation function to obtain the motion control quantity;

[0039] The third acquisition module is used to input the motion control quantity into the performance evaluation application approximation function to obtain the motion controller performance evaluation index.

[0040] The control module is used to optimize the motion control application approximation function according to the motion controller performance evaluation index, so as to control the vehicle according to the motion control quantity output by the optimized motion control application approximation function;

[0041] The update module is used to perform online updates based on a performance evaluation training approximation function with the same parameters as the performance evaluation application approximation function, and / or to update the motion control application approximation function online based on a motion control training approximation function with the same parameters as the motion control application approximation function and the performance evaluation training approximation function.

[0042] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the vehicle motion control method described in any of the preceding claims.

[0043] The present invention also provides an autonomous driving vehicle, including the electronic equipment described above.

[0044] According to the present invention, an autonomous vehicle includes an articulated truck;

[0045] The driving data of the articulated truck also includes at least one of the following: front and rear body articulation angle, front and rear body articulation angular velocity, rear body roll angle, rear body roll angular velocity, and rear body yaw angle.

[0046] The performance evaluation index values ​​of the articulated truck also include at least one of the following: the oscillation frequency of the front and rear vehicle body hinge angles, the oscillation decay rate of the front and rear vehicle body hinge angles, and the rear vehicle body rollover threshold.

[0047] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the vehicle motion control method described in any of the preceding claims.

[0048] This invention provides a vehicle motion control method, device, electronic device, vehicle, and storage medium. The method involves acquiring real-time vehicle driving data; inputting the real-time driving data into a motion control application approximation function to obtain a motion control quantity; inputting the motion control quantity into a performance evaluation application approximation function to obtain a motion controller performance evaluation index; optimizing the motion control application approximation function based on the motion controller performance evaluation index; and controlling the vehicle based on the motion control quantity output by the optimized motion control application approximation function. The performance evaluation application approximation function is updated online based on a performance evaluation training approximation function with the same parameters, and / or the motion control application approximation function is updated online based on both a motion control training approximation function with the same parameters and a performance evaluation training approximation function. This allows for the gradual learning of a motion control method suitable for the real vehicle operating environment using selected real driving data, even without prior information about an accurate vehicle dynamics model. Furthermore, the control model is updated online using a function model with the same parameters as the control model and a performance evaluation model, ensuring continuous optimization of the motion control method. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0050] Figure 1 This is one of the flowcharts of the vehicle motion control method provided by the present invention;

[0051] Figure 2 This is the second flowchart of the vehicle motion control method provided by the present invention;

[0052] Figure 3 This is the third flowchart of the vehicle motion control method provided by the present invention;

[0053] Figure 4 This is the fourth flowchart of the vehicle motion control method provided by the present invention;

[0054] Figure 5 This is a schematic diagram of the structure of a vehicle motion control device provided by the present invention;

[0055] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0057] Figure 1 A flowchart of the vehicle motion control method provided in the embodiments of the present invention is shown below. Figure 1 As shown, the vehicle motion control method provided in this embodiment of the invention includes:

[0058] Step 101: Obtain real-time vehicle driving data;

[0059] Step 102: Input the real-time vehicle driving data into the motion control application approximation function to obtain the motion control quantity;

[0060] Step 103: Input the motion control quantity into the performance evaluation application approximation function to obtain the motion controller performance evaluation index; wherein, the performance evaluation application approximation function is updated online based on the performance evaluation training approximation function with the same parameters.

[0061] Step 104: Optimize the motion control application approximation function based on the motion controller performance evaluation index, so as to control the vehicle based on the motion control quantity output by the optimized motion control application approximation function; the motion control application approximation function is updated online based on the motion control training approximation function with the same parameters and the performance evaluation training approximation function.

[0062] In this embodiment of the invention, the approximation function includes polynomial functions, lookup table functions, neural networks, etc., and this application does not limit the selection of the approximation function.

[0063] In some embodiments of the present invention, the approximate function applied for performance evaluation and the approximate function applied for motion control are iteratively updated alternately at preset intervals, so as to ensure the optimal control performance of the motion controller while ensuring system stability.

[0064] Traditional autonomous driving motion control algorithms mainly include expert system control, PID control, linear quadratic programming, and model predictive control. These algorithms are designed based on expert experience rules or simplified vehicle models. Their performance is highly dependent on the completeness of the experience rules and the accuracy of the vehicle model control. Once deployed to the onboard controller, the vehicle model algorithm cannot be updated or optimized, and its accuracy cannot be continuously improved. Furthermore, because these algorithms are designed for conventional operating conditions, their control performance is poor for special operating conditions encountered by articulated trucks in actual operation, such as "frequent braking" and "dragging," making it difficult to meet the road traffic safety requirements of autonomous articulated trucks during high-speed long-distance transportation.

[0065] This invention provides a vehicle motion control method that acquires real-time vehicle driving data; inputs the real-time vehicle driving data into a motion control application approximation function to obtain a motion control quantity; inputs the motion control quantity into a performance evaluation application approximation function to obtain a motion controller performance evaluation index; optimizes the motion control application approximation function based on the motion controller performance evaluation index, and controls the vehicle based on the motion control quantity output by the optimized motion control application approximation function; wherein, the performance evaluation application approximation function is updated online based on a performance evaluation training approximation function with the same parameters, and / or, the motion control application approximation function is updated online based on both a motion control training approximation function with the same parameters and a performance evaluation training approximation function. This method allows for the gradual learning of a motion control method suitable for the real vehicle operating environment using selected real driving data, even without prior information on an accurate vehicle dynamics model. Furthermore, by updating the control model online using a function model with the same parameters as the control model and a performance evaluation model, the motion control method is continuously optimized, meeting the road traffic safety requirements of autonomous vehicles, especially articulated trucks, during high-speed long-distance transportation.

[0066] Based on any of the above embodiments, such as Figure 2 As shown, the performance evaluation application approximation function provided in this embodiment of the invention is updated online based on the performance evaluation training approximation function with the same parameters, including:

[0067] Step 201: Obtain the training data of the performance evaluation training approximation function from the training database. The training data of the performance evaluation training approximation function includes the training data of the motion control quantity and its corresponding motion control evaluation index label value.

[0068] In this embodiment of the invention, the motion control training data includes, but is not limited to, the vehicle's future trajectory data and the vehicle's acceleration data at the next moment.

[0069] Step 202: Input the motion control quantity training data into the performance evaluation training approximation function to obtain the predicted value of the motion controller performance evaluation index;

[0070] In this embodiment of the invention, the method for obtaining the predicted value of the motion controller performance evaluation index specifically includes:

[0071] Step 2021: Input the vehicle's acceleration data at the next moment into the performance evaluation training approximation function to obtain the first motion control evaluation index, which is the deviation between the vehicle's speed and the target speed;

[0072] Step 2022: Input the future trajectory data of the vehicle into the performance evaluation training approximation function to obtain the second motion control evaluation index. The second motion control evaluation index is the deviation between the future trajectory of the vehicle output by the motion controller and the actual future trajectory of the vehicle during operation.

[0073] Step 2023: Normalize the first motion control evaluation index and the second motion control evaluation index, then sum them up, and select the cumulative value of the summation over a period of time as the predicted value of the motion controller performance evaluation index.

[0074] Step 203: Solve the loss function of the performance evaluation training approximation function using the gradient descent method to update the parameters of the performance evaluation training approximation function. The loss function of the performance evaluation training approximation function is constructed based on the predicted value of the motion controller performance evaluation index and the labeled value of the motion control evaluation index.

[0075] Step 204: After reducing the parameters of the updated performance evaluation training approximation function by a preset ratio, assign them to the performance evaluation application approximation function to update the performance evaluation application approximation function online.

[0076] Since the computing power of the vehicle controller is limited, the algorithm model deployed inside it cannot be automatically iterated and updated. In this embodiment of the invention, by directly assigning the updated parameters of the performance evaluation training approximation function to the performance evaluation application approximation function, the online update of the performance evaluation application approximation function in the vehicle controller can be realized, ensuring the optimality of the motion control algorithm in the vehicle controller.

[0077] Based on any of the above embodiments, such as Figure 3As shown, the motion control application approximation function provided in this embodiment of the invention is updated online based on the motion control training approximation function and the performance evaluation training approximation function with the same parameters, including:

[0078] Step 301: Obtain training driving data from the training database and input the training driving data into the motion control training approximation function to obtain motion control quantity training data;

[0079] The aforementioned training driving data includes historical trajectory data of the vehicle itself, historical trajectory data of surrounding vehicles, and future trajectory data of surrounding vehicles obtained from the collected data. The collected data includes, but is not limited to, the vehicle's speed, current acceleration, yaw rate, position information of the vehicle itself, and position information of surrounding vehicles. This data is data that can be directly obtained from existing vehicles. Based on other prediction modules on the vehicle, the vehicle's dynamics information, positioning information, and positioning information of surrounding vehicles can be obtained and selected as the original information for the design of its motion control algorithm, serving as the basis for subsequent optimal motion control.

[0080] Step 302: Calculate the gradient of the performance evaluation training approximation function on the motion control quantity training data to obtain the performance evaluation gradient;

[0081] Step 303: Multiply the performance evaluation gradient with the original motion control gradient of the motion control training approximation function to obtain the motion control gradient of the motion control training approximation function;

[0082] Step 304: Solve the loss function of the motion control training approximation function based on the motion control gradient of the motion control training approximation function, so as to update the parameters of the motion control training approximation function. The loss function of the motion control training approximation function is constructed based on the motion control quantity training data and the motion control quantity labeled data.

[0083] Step 305: After scaling down the parameters of the updated motion control training approximation function according to a preset ratio, assign the values ​​to the motion control application approximation function to update the motion control application approximation function online.

[0084] Since the computing power of the vehicle controller is limited, the algorithm model deployed inside it cannot be automatically iterated and updated. In this embodiment of the invention, by directly assigning the updated parameters of the motion control training approximation function to the motion control application approximation function, the motion control application approximation function in the vehicle controller can be updated online, ensuring the optimality of the motion control algorithm in the vehicle controller.

[0085] Based on any of the above embodiments, such as Figure 4 As shown, the method for obtaining training data provided in this embodiment of the invention includes:

[0086] Step 401: Collect vehicle operation data under different driving conditions such as different driving sections, different driving speeds, and different traffic volumes;

[0087] Step 402: Recombine the vehicle operation data to obtain at least one data combination; the data combination includes the vehicle's historical trajectory data combination, the vehicle's weekly historical trajectory data combination, the vehicle's future trajectory data combination, the vehicle's speed data combination, the target speed data combination, the acceleration data combination, and the vehicle's weekly future trajectory data combination.

[0088] Step 403: Align the data from different data combinations with the same timestamp to obtain training driving data and motion control quantity training data;

[0089] Step 404: Construct a training database based on the training driving data and its corresponding motion control quantity annotation data, as well as the motion control quantity training data and its corresponding motion control evaluation index annotation values.

[0090] In this embodiment of the invention, the easily obtainable vehicle dynamics information, vehicle positioning information, and surrounding vehicle positioning information are selected as the original information for the motion control algorithm design, and these information are recombined and used as the basis for subsequent optimal motion control.

[0091] In some embodiments of the invention, the autonomous vehicle is an articulated truck. The driving data of the articulated truck also includes the front and rear body articulation angles, the front and rear body articulation angular velocities, the rear body roll angle, the rear body roll angular velocity, and the rear body yaw angle. The driving data is input into the motion control training approximation function to obtain the predicted values ​​of the motion control parameters. For example, the performance evaluation index values ​​of the articulated truck also include the oscillation frequency of the front and rear body articulation angles, the oscillation decay rate of the front and rear body articulation angles, and the rear body rollover threshold.

[0092] Since the steering control angle of an articulated truck needs to take into account information such as the current front and rear vehicle hinge angles and hinge angular velocities, and motion control needs to prevent the vehicle from rolling over, it is more complex than other passenger vehicles. The vehicle motion control method provided in this embodiment of the invention is more suitable for motion control in real driving environments. It has strong applicability to articulated trucks in dynamic traffic environments and has high computational efficiency.

[0093] The following describes a vehicle motion control device provided by the present invention. The vehicle motion control device described below can be referred to in correspondence with the vehicle motion control method described above.

[0094] Figure 5 This is a schematic diagram of a vehicle motion control device provided in an embodiment of the present invention, such as... Figure 5 As shown, the vehicle motion control device provided in this embodiment of the invention includes:

[0095] The first acquisition module 501 is used to acquire real-time vehicle driving data;

[0096] The second acquisition module 502 is used to input real-time vehicle driving data into the motion control application approximation function to obtain motion control quantities.

[0097] The third acquisition module 503 is used to input the motion control quantity into the performance evaluation application approximation function to obtain the motion controller performance evaluation index.

[0098] Control module 504 is used to optimize the motion control application approximation function according to the motion controller performance evaluation index, so as to control the vehicle according to the motion control quantity output by the optimized motion control application approximation function;

[0099] The update module 505 is used to perform online updates based on the performance evaluation training approximation function with the same parameters as the performance evaluation application approximation function, and / or to update the motion control application approximation function online based on the motion control training approximation function with the same parameters as the motion control application approximation function and the performance evaluation training approximation function.

[0100] This invention provides a vehicle motion control device that acquires real-time vehicle driving data; inputs the real-time vehicle driving data into a motion control application approximation function to obtain a motion control quantity; inputs the motion control quantity into a performance evaluation application approximation function to obtain a motion controller performance evaluation index; optimizes the motion control application approximation function based on the motion controller performance evaluation index, and controls the vehicle based on the motion control quantity output by the optimized motion control application approximation function; wherein, the performance evaluation application approximation function is updated online based on a performance evaluation training approximation function with the same parameters, and / or, the motion control application approximation function is updated online based on both a motion control training approximation function with the same parameters and a performance evaluation training approximation function. This allows for the gradual learning of a motion control method suitable for the real vehicle operating environment using selected real driving data, even without prior information on an accurate vehicle dynamics model. Furthermore, by updating the control model online using a function model with the same parameters as the control model and a performance evaluation model, the continuous optimization of the motion control method is ensured.

[0101] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6As shown, the electronic device may include a processor 610, a communication interface 620, a memory 630, and a communication bus 640, wherein the processor 610, the communication interface 620, and the memory 630 communicate with each other via the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute a vehicle motion control method, which includes: acquiring real-time vehicle driving data; inputting the real-time vehicle driving data into a motion control application approximation function to obtain a motion control quantity; inputting the motion control quantity into a performance evaluation application approximation function to obtain a motion controller performance evaluation index; optimizing the motion control application approximation function based on the motion controller performance evaluation index, so as to control the vehicle based on the motion control quantity output by the optimized motion control application approximation function; wherein the performance evaluation application approximation function is updated online based on a performance evaluation training approximation function with the same parameters, and / or, the motion control application approximation function is updated online based on both a motion control training approximation function with the same parameters and a performance evaluation training approximation function.

[0102] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0103] This invention also provides an autonomous driving vehicle, including the electronic equipment described in the above embodiments.

[0104] In this embodiment of the invention, the autonomous vehicle includes an articulated truck.

[0105] The driving data of the articulated truck also includes at least one of the following: front and rear body articulation angle, front and rear body articulation angular velocity, rear body roll angle, rear body roll angular velocity, and rear body yaw angle.

[0106] The performance evaluation index values ​​of the articulated truck also include at least one of the following: the oscillation frequency of the front and rear vehicle body hinge angles, the oscillation decay rate of the front and rear vehicle body hinge angles, and the rear vehicle body rollover threshold.

[0107] On the other hand, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the vehicle motion control method provided by the above methods. The method includes: acquiring real-time vehicle driving data; inputting the real-time vehicle driving data into a motion control application approximation function to obtain a motion control quantity; inputting the motion control quantity into a performance evaluation application approximation function to obtain a motion controller performance evaluation index; optimizing the motion control application approximation function based on the motion controller performance evaluation index, so as to control the vehicle based on the motion control quantity output by the optimized motion control application approximation function; wherein the performance evaluation application approximation function is updated online based on a performance evaluation training approximation function with the same parameters, and / or, the motion control application approximation function is updated online based on both a motion control training approximation function with the same parameters and a performance evaluation training approximation function.

[0108] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0109] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A vehicle motion control method, characterized in that, include: Obtain real-time vehicle driving data; The real-time driving data of the vehicle is input into the approximate function of the motion control application to obtain the motion control quantity; The motion control quantity is input into the performance evaluation application approximation function to obtain the motion controller performance evaluation index; The motion control application approximation function is optimized based on the motion controller performance evaluation index, so as to control the vehicle based on the motion control quantity output by the optimized motion control application approximation function; Wherein, the performance evaluation application approximation function is updated online based on the performance evaluation training approximation function with the same parameters, and / or, the motion control application approximation function is updated online based on the motion control training approximation function with the same parameters and the performance evaluation training approximation function; The motion control application approximation function is updated online based on the motion control training approximation function and the performance evaluation training approximation function with the same parameters, including: The training driving data is obtained from the training database and input into the motion control training approximation function to obtain motion control quantity training data. The performance evaluation training approximation function is used to calculate the gradient of the motion control quantity training data to obtain the performance evaluation gradient. Multiply the performance evaluation gradient by the original motion control gradient of the motion control training approximation function to obtain the motion control gradient of the motion control training approximation function. The loss function of the motion control training approximation function is solved based on the motion control gradient of the motion control training approximation function to update the parameters of the motion control training approximation function. The loss function of the motion control training approximation function is constructed based on the motion control quantity training data and the motion control quantity labeled data. The parameters of the updated motion control training approximation function are reduced by a preset ratio and then assigned to the motion control application approximation function to update the motion control application approximation function online.

2. The vehicle motion control method according to claim 1, characterized in that, The performance evaluation application approximation function is updated online based on the performance evaluation training approximation function with the same parameters, specifically including: Obtain performance evaluation training approximation function training data from the training database. The performance evaluation training approximation function training data includes motion control quantity training data and its corresponding motion control evaluation index labeled values. The motion control quantity training data is input into the performance evaluation training approximation function to obtain the predicted value of the motion controller performance evaluation index. The loss function of the performance evaluation training approximation function is solved using the gradient descent method to update the parameters of the performance evaluation training approximation function. The loss function of the performance evaluation training approximation function is constructed based on the predicted value of the motion controller performance evaluation index and the labeled value of the motion control evaluation index. The parameters of the updated performance evaluation training approximation function are reduced by a preset ratio and then assigned to the performance evaluation application approximation function to update the performance evaluation application approximation function online.

3. The vehicle motion control method according to claim 2, characterized in that, The motor control training data includes: The vehicle's future trajectory data and acceleration data at the next moment.

4. The vehicle motion control method according to claim 3, characterized in that, The method for obtaining the predicted values ​​of the motion controller performance evaluation indicators includes: The acceleration data of the vehicle at the next moment is input into the performance evaluation training approximation function to obtain the first motion control evaluation index, which is the deviation between the vehicle speed and the target speed. The vehicle's future trajectory data is input into the performance evaluation training approximation function to obtain a second motion control evaluation index. The second motion control evaluation index is the deviation between the vehicle's future trajectory output by the motion controller and the actual future trajectory of the vehicle during operation. The first motion control evaluation index and the second motion control evaluation index are normalized and then summed. The cumulative value of the summation over a period of time is selected as the predicted value of the motion controller performance evaluation index.

5. The vehicle motion control method according to claim 1, characterized in that, The training driving data includes: The historical trajectory data of the vehicle, the historical trajectory data of the surrounding vehicle, and the future trajectory data of the surrounding vehicle obtained from the collected data include at least one of the following: the vehicle's speed, the vehicle's current acceleration, the yaw rate, the vehicle's position information, and the surrounding vehicle's position information.

6. A vehicle motion control method according to claim 1 or 2, characterized in that, The method for establishing the training database includes: Collect vehicle driving data under different driving conditions, including different driving road segments, different driving speeds, and different traffic volumes. The vehicle driving data is recombined to obtain at least one data combination; the data combination includes the vehicle's historical trajectory data combination, the vehicle's weekly historical trajectory data combination, the vehicle's future trajectory data combination, the vehicle's speed data combination, the target speed data combination, the acceleration data combination, and the vehicle's weekly future trajectory data combination. Data from different data combinations with a unified timestamp are aligned to obtain training driving data and motion control quantity training data. A training database is constructed based on the training driving data and its corresponding motion control quantity annotation data, as well as the motion control quantity training data and its corresponding motion control evaluation index annotation values.

7. The vehicle motion control method according to claim 2, characterized in that, When both the performance evaluation application approximation function and the motion control application approximation function are updated online, the following is included: The performance evaluation application uses an approximation function, and the motion control application uses an approximation function, which is iteratively updated alternately at preset intervals.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the vehicle motion control method as described in any one of claims 1 to 7.

9. An autonomous driving vehicle, characterized in that, Including the electronic device as described in claim 8.

Images