Vehicle chassis cyber-physical system and multi-agent collaborative control method

Abstract

The application provides a vehicle chassis information physical system and a multi-agent collaborative control method, comprising a perception module, a multi-agent coordination control module, a line control chassis physical system, a driving environment digital twin and a vehicle model digital twin; the application is based on the multi-agent coordination control module, coordinates each upper-layer automobile safety auxiliary system of the automobile chassis, avoids the conflict behaviors between the control systems, at the same time, reduces the decision difficulty of the lower-layer line control system, and improves the chassis domain comprehensive collaborative control performance under specific working conditions. Based on the digital twin method, the driving environment digital twin and the vehicle model digital twin are established, the digital twin and the real environment interact in real time, the control quantity calculated based on the vehicle model digital twin is more accurate than the conventional method, and the stability and reliability of the chassis control are greatly improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of automobile drive-by-wire chassis system, and particularly relates to a vehicle chassis cyber-physical system and a multi-agent collaborative control method. BACKGROUND

[0002] With the transformation and upgrading of China from a large automobile country to a powerful automobile country, the intelligent automobile field representing the cutting-edge technology in the automobile field is booming, and the drive-by-wire chassis system as a key system of intelligent automobiles has become a research hotspot in recent years.

[0003] The drive-by-wire chassis system includes a drive-by-wire braking subsystem, a drive-by-wire steering subsystem, a drive-by-wire suspension subsystem, a drive-by-wire driving subsystem and various upper automobile safety auxiliary control systems. At present, most of the researches on chassis collaborative control focus on the collaborative control among the drive-by-wire braking subsystem, the drive-by-wire steering subsystem and the drive-by-wire suspension subsystem. The control method of a cloud control intelligent chassis system based on multi-agent is disclosed in the invention patent with the application number 202110225610.1. In the patent, the drive-by-wire braking subsystem, the drive-by-wire steering subsystem and the drive-by-wire suspension subsystem are regarded as agents, the outputs of the three are coordinated through a multi-agent coordinator, and the multi-agent reinforcement learning results are regularly checked and adjusted through a cloud processing center. The invention still directly coordinates and controls the drive-by-wire braking subsystem, the drive-by-wire steering subsystem and the drive-by-wire suspension subsystem, which are lower control systems, and ignores the control target differences among the upper automobile safety auxiliary systems when the automobile safety auxiliary systems as the upper controllers of the drive-by-wire systems play a role in special working conditions. Directly coordinating and controlling the lower control systems may easily lead to decision conflicts between the upper and lower systems, coordination difficulties of the lower drive-by-wire subsystems and complex comprehensive decision-making, and it is difficult to achieve the optimal control state in some working conditions. Therefore, in order to reduce the decision difficulty of the lower drive-by-wire systems and improve the comprehensive coordination performance of the chassis domain in special working conditions, it is necessary to coordinate and control the upper safety auxiliary systems.

[0004] The vehicle reference model used in the chassis-by-wire technology is generally a simplified vehicle mathematical model. The simplified vehicle mathematical model after simplifying the degrees of freedom of the vehicle can achieve the purpose of rapid calculation. However, the vehicle is actually a nonlinear model, which is difficult to express by an analytical expression. Too much simplification of the vehicle physical model results in a large error between the vehicle reference model and the actual state of the vehicle, which leads to inaccurate control parameters calculated based on the reference model, and it is difficult to achieve precise motion control of the vehicle. At present, the method based on digital twinning technology is introduced to solve this problem. The invention patent with the application number 202110170344.7 discloses an intelligent chassis-by-wire system driven by digital twinning and a fault diagnosis method thereof. The patent establishes a chassis-by-wire steering system model, a chassis-by-wire braking system model and a chassis-by-wire driving system model in the chassis twinning system, and achieves the purpose of information exchange by using the real-time interaction between the chassis twinning system and the chassis-by-wire device. However, this invention only focuses on the chassis state and does not further consider the dynamic change process of the chassis twinning body in the real-time driving environment. In addition, it also does not consider the influence of the chassis-by-wire suspension system on the dynamic characteristics of the chassis. The above problems will lead to an inaccurate chassis twinning body model and only very limited vehicle parameters can be obtained. Therefore, it is necessary to establish a real-time driving environment twinning body to further improve the model accuracy of the chassis twinning body and obtain more vehicle parameters. SUMMARY

[0005] In view of the above technical problems, the present application provides a vehicle chassis cyber-physical system and a multi-agent collaborative control method, which solves the problems of insufficient accuracy of the vehicle reference model and coordination of the upper-layer vehicle safety auxiliary system of the chassis system, and improves the comprehensive collaborative control performance of the chassis-by-wire system.

[0006] The vehicle chassis cyber-physical system of the present application comprises a perception module, a multi-agent coordination control module, a chassis-by-wire physical system, a driving environment digital twinning body and a vehicle model digital twinning body. Based on the multi-agent coordination control module, the present application coordinates the upper-layer vehicle safety auxiliary system of the vehicle chassis, avoids the conflict behavior between the control systems, reduces the decision difficulty of the lower-layer chassis-by-wire system, and improves the comprehensive collaborative control performance of the chassis domain under specific working conditions. Based on the digital twinning method, the driving environment digital twinning body and the vehicle model digital twinning body are established. The twinning body interacts with the real environment in real time, and the control quantity calculated based on the vehicle model digital twinning body is more accurate than that obtained by the conventional method, which greatly improves the stability and reliability of the chassis control.

[0007] Based on the digital twinning method and the multi-agent, the present application comprehensively constructs the chassis physical layer, the network layer, the control flow and the information flow into a vehicle chassis cyber-physical system, which improves the comprehensive collaborative control performance of the chassis and has important significance for improving the comfort, safety and smoothness of the whole vehicle.

[0008] Note that the description of these objects does not preclude the presence of other objects. One embodiment of the present application does not need to achieve all the above objects. The objects other than the above can be extracted from the description, drawings, claims.

[0009] The present application achieves the above technical objects through the following technical means.

[0010] A vehicle chassis information physical system comprises a perception module, a multi-agent coordinated control module, a by-wire chassis physical system, a driving environment digital twin and a vehicle model digital twin.

[0011] The perception module is connected with vehicle-mounted sensors for collecting vehicle state information and environmental information; the perception module is also connected with the driving environment digital twin for transmitting real-time driving environment to the driving environment digital twin to build a virtual driving environment that is real-time synchronized with the real driving environment.

[0012] The by-wire chassis physical system is used to execute the control instructions of the multi-agent coordinated control module.

[0013] The driving environment digital twin is provided with a global map, and the environmental information is mapped into the driving environment digital twin in real time through the perception module to obtain a virtual real-time driving environment consistent with the real driving environment.

[0014] The vehicle model digital twin interacts with the real by-wire chassis physical system in real time to map the driving state of the actual vehicle in the actual driving environment in real time, and obtain more accurate and rich vehicle state parameters to provide a more accurate vehicle reference model for chassis control.

[0015] The multi-agent coordinated control module selects and calls each automobile safety auxiliary system to control the by-wire chassis physical system in combination with the current driving condition and driving environment, and obtains the best action to realize the decision intention through reinforcement learning, so as to coordinate the outputs of each automobile safety auxiliary system and comprehensively coordinate and control the vehicle.

[0016] In the above scheme, the by-wire chassis physical system comprises a by-wire steering subsystem, a wheel system, a by-wire brake subsystem, a by-wire suspension subsystem and a by-wire drive subsystem; the vehicle model digital twin comprises a by-wire drive twin, a by-wire suspension twin, a by-wire brake twin, a by-wire steering twin and four wheel twins.

[0017] In the above scheme, the multi-agent coordinated control module comprises an upper vehicle safety auxiliary system, a coordination agent and a decision layer; the coordination agent selects and calls each upper vehicle safety auxiliary system according to the requirements of the decision layer and coordinates the vehicle control to make the vehicle comprehensive control state optimal.

[0018] In the scheme, the vehicle safety auxiliary system includes an emergency braking system, a vehicle body stability system, an anti-lock braking system, a front wheel active steering system, a four-wheel steering system, an intelligent vehicle comprehensive auxiliary steering system, a four-wheel drive system, an adaptive cruise system, an active suspension system, and a driving force control system.

[0019] A control method of the vehicle chassis information physical system comprises the following steps:

[0020] 1) The perception module collects driving environment information and road information and transmits them to the driving environment twin model, collects vehicle state information and driver instruction information and transmits them to the multi-agent coordination control module;

[0021] 2) The driving environment twin establishes a local map in the current driving environment in real time according to the received information from the perception module, combines the existing global map, and constructs a virtual real-time driving environment;

[0022] 3) The multi-agent coordination control module receives information from the perception module and the vehicle model digital twin, decides to call the vehicle safety auxiliary system, coordinates the outputs of the vehicle safety auxiliary systems through the coordination agent, and transmits the optimal output control quantity to the drive-by-wire chassis physical system;

[0023] 4) The drive-by-wire chassis physical system receives the control instruction from the multi-agent coordination control module, executes the control instruction through the driving actuator, and transmits the chassis state information to the vehicle model digital twin;

[0024] 5) The vehicle model digital twin updates and optimizes the vehicle model digital twin in real time according to the received information from the drive-by-wire chassis physical system, combines the existing virtual vehicle model, and interacts with the driving environment digital twin in real time, truly maps the driving state of the vehicle in the actual driving environment, and transmits the running state information of the twin vehicle model to the multi-agent coordination control module.

[0025] In the scheme, the perception module information in step 1) includes external environment information, vehicle state information, and driver control instruction information; the external environment information is measured by a laser radar and a depth camera, and data fusion is performed in the perception module; the vehicle state information is measured by a vehicle-mounted sensor, and state estimation and filtering processing are performed in the perception module.

[0026] In the scheme, the driving environment twin model in step 2) is set in a cloud server, transmits information with the vehicle-mounted perception module through a 5G cellular network, matches the perception information with the built-in global map, and constructs a high-precision virtual driving environment.

[0027] In the above scheme, the specific working process of the multi-agent coordination control module in step 3) includes:

[0028] 31) The perception module collects the current driving environment information, the vehicle state information and the driver instruction information, and transmits them to the decision layer of the multi-agent coordination control module;

[0029] 32) The decision layer first combines the vehicle state and the driving environment information to judge the current state stability of the vehicle and predict the state stability in a preset time in the future, if it is judged that the vehicle state will remain stable and the driver is in the driving mode, the multi-agent coordination control module directly transmits the information stream to the physical system of the drive-by-wire chassis;

[0030] 33) If it is judged that the vehicle state will remain stable and the driver is in the automatic driving mode, the decision layer selects to call one or more vehicle safety auxiliary systems with the goal of maintaining the stable and safe driving of the vehicle, and the coordination agent coordinates the outputs of the vehicle safety auxiliary systems;

[0031] 34) If it is judged that the vehicle state is unstable, the decision layer selects to call one or more vehicle safety auxiliary systems with the goal of maintaining the stability of the vehicle state in combination with the vehicle state and the driving environment information, and the coordination agent coordinates the outputs of the vehicle safety auxiliary systems.

[0032] In the above scheme, the chassis state information in step 4) is transmitted to the vehicle model digital twin located on the cloud server through the 5G cellular network, and the twin vehicle model is continuously corrected based on the real-time state of the real vehicle chassis.

[0033] In the above scheme, the vehicle model digital twin in step 5) is placed in the cloud server, and the twin vehicle model runs in the constructed twin driving environment, which is a real-time mapping of the vehicle running in the real driving environment.

[0034] Compared with the prior art, the beneficial effects of the present application are:

[0035] 1. Based on the characteristics of the drive-by-wire chassis physical layer, network layer and control layer, the present application combines digital twin technology to construct a chassis information physical system architecture, greatly improves the comprehensive control ability of the chassis domain, and at the same time establishes a driving environment digital twin and a vehicle model digital twin, which obtains an accurate vehicle reference model through real-time interaction between virtual driving and real driving, and has important significance for improving the control accuracy of the chassis.

[0036] 2. The application is based on multi-agent theory, fully considers the control target differences between various vehicle safety auxiliary systems as upper controllers of each drive-by-wire subsystem under certain working conditions, cooperatively controls the outputs of the upper vehicle safety auxiliary systems, reduces the decision difficulty of the lower drive-by-wire system, improves the comprehensive cooperation performance of the chassis domain under special working conditions, and has important significance for improving the comfort, safety and smoothness of the vehicle.

[0037] Note that the description of these effects does not preclude the existence of other effects. One embodiment of the present application does not necessarily have all the above-mentioned effects. Effects other than the above-mentioned effects can be clearly seen and extracted from the description, drawings, claims, etc. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 The chassis cyber-physical system overall architecture diagram of an embodiment of the present application.

[0039] Figure 2 The internal structure diagram of each module of the chassis cyber-physical system of an embodiment of the present application.

[0040] Figure 3 The multi-agent coordination control principle diagram of an embodiment of the present application.

[0041] Figure 4 The common driving lane change working condition scene diagram of an embodiment of the present application.

[0042] Figure 5 The multi-agent coordination control architecture diagram of an embodiment of the present application.

[0043] Figure 6 The common driving lane change working condition workflow diagram of an embodiment of the present application.

[0044] Figure 7 The multi-agent coordination control architecture diagram of a low adhesion road over-bend stability control working condition of a special application scenario one of the present application.

[0045] Figure 8 The multi-agent coordination control result diagram of a low adhesion road over-bend stability control working condition of a special application scenario one of the present application.

[0046] Figure 9 The multi-agent coordination control architecture diagram of a driver distraction human-machine co-driving control working condition of a special application scenario two of the present application.

[0047] Figure 10A coordination control result diagram of a multi-agent in a driver distraction human-machine co-driving control condition for the second special application scenario of the application. DETAILED DESCRIPTION

[0048] Embodiments of the application are described below in detail, examples of which are shown in the drawings, wherein the same or similar reference numbers represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by reference to the drawings are exemplary and are intended to explain the application, and cannot be understood as a limitation of the application.

[0049] Figure 1 A preferred embodiment of the vehicle chassis cyber-physical system architecture is shown, which includes a perception module, a multi-agent coordination control module, a drive-by-wire chassis physical system, a driving environment digital twin, and a vehicle model digital twin, and the internal structure of each module of the system is shown as Figure 2

[0050] The perception module is connected with all vehicle-mounted sensors for collecting vehicle state information and driving environment information, and is connected with the driving environment digital twin for transmitting real-time driving environment to the driving environment digital twin to build a virtual driving environment that is real-time synchronized with the real driving environment. The information collected by the perception module includes: driving environment image information, driving environment three-dimensional point cloud information, road curvature, road roughness, vehicle lateral, longitudinal, and vertical speed and acceleration, wheel speed and slip ratio, vehicle body yaw angular velocity, mass center side slip angle, driving torque, braking torque, suspension damping force, suspension stiffness, suspension deflection, suspension load, and driver instruction information.

[0051] The multi-agent coordination control module includes an upper vehicle safety auxiliary system and a coordination agent, which selects and calls each upper vehicle safety auxiliary system according to the decision-making requirements of the coordination agent and coordinates the vehicle control.

[0052] The upper vehicle safety auxiliary system includes AEB (emergency braking system), ESP (vehicle body stability system), ABS (anti-lock braking system), AFS (front wheel active steering system), 4WS (four-wheel steering system), EPS (intelligent vehicle comprehensive auxiliary steering system), 4WD (four-wheel steering system), ACC (adaptive cruise control system), ASS (active suspension system), and TCS (drive power control system) and other vehicle safety auxiliary systems. The EPS system is a comprehensive auxiliary steering system based on an intelligent vehicle, which has functions in addition to traditional auxiliary steering, including driving mode judgment, guided steering, etc.

[0053] The coordination agent is used to coordinate the control output of the upper vehicle safety auxiliary system to make the vehicle comprehensive control state optimal.​

[0054] The drive-by-wire chassis physical system comprises a drive-by-wire steering subsystem, a drive-by-wire braking subsystem, a drive-by-wire suspension subsystem and a drive-by-wire driving subsystem, which are used to execute the control instructions of the multi-agent coordination control module.

[0055] The drive environment digital twin has a global map, and the environment information is mapped into the drive environment digital twin in real time through the perception module, and is matched with the global map in real time, so as to obtain a virtual real-time driving environment consistent with the real driving environment.

[0056] The vehicle model digital twin comprises a drive-by-wire driving twin, a drive-by-wire suspension twin, a drive-by-wire braking twin, a drive-by-wire steering twin and four wheel twins.

[0057] The drive-by-wire driving twin is derived by real-time mapping of the chassis physical drive-by-wire driving subsystem to the digital model, and its characteristics are consistent with the physical drive-by-wire driving system, and it interacts with the physical system in real time to correct the twin model. The drive-by-wire suspension twin is derived by real-time mapping of the chassis physical drive-by-wire suspension subsystem to the digital model, and its characteristics are consistent with the physical drive-by-wire suspension system, and it interacts with the physical system in real time to correct the twin model. The drive-by-wire braking twin is derived by real-time mapping of the chassis physical drive-by-wire braking subsystem to the digital model, and its characteristics are consistent with the physical drive-by-wire braking system, and it interacts with the physical system in real time to correct the twin model. The drive-by-wire steering twin is derived by real-time mapping of the chassis physical drive-by-wire steering subsystem to the digital model, and its characteristics are consistent with the physical drive-by-wire steering system, and it interacts with the physical system in real time to correct the twin model.

[0058] The wheel twin is derived by real-time mapping of the physical wheel to the digital model, and its characteristics are consistent with the physical wheel, and it interacts with the physical wheel system in real time to correct the twin model.

[0059] The multi-agent coordination control module selects and calls the upper vehicle safety auxiliary system to control the chassis in combination with the current driving condition and driving environment, and obtains the best action to realize the decision intention through reinforcement learning, so as to coordinate the outputs of the vehicle safety auxiliary system controllers and comprehensively coordinate and control the vehicle.

[0060] The specific control process of the multi-agent coordination control module is that after receiving the information transmitted by the perception module, the decision layer comprehensively judges whether the vehicle state is safe, whether the vehicle state is safe in the future period of time and whether the driver's intention will cause the vehicle to lose stability in combination with the current driving environment information, the vehicle state information and the driver's instruction information, makes a comprehensive decision, selects and calls the upper vehicle safety auxiliary system, and transmits the coordinated best control amount to the lower execution subsystem through the coordination of the intelligent agent.

[0061] The principle of the multi-agent coordination control strategy is as follows:Figure 3 As shown, the agent interacts with the external environment. When the environment is in state s, the agent observes the environment state as observation value o. Based on the observed environment state, the agent makes a decision according to the policy function π (a probability density function), controlling the agent to perform the corresponding action a. For each action performed, a reward is given by the agent. The goal of reinforcement learning is to maximize the total reward. After performing the current action, the agent enters the next state, described by a state transition function determined by the environment. Furthermore, the action value function Q is used to evaluate the quality of the action decided by the agent based on the policy function π in a specific state. The policy function π that maximizes the action value function Q is selected; this process is the update process of the π function.

[0062] The drive-by-wire chassis physical system is also connected to a multi-agent coordination control module to execute the output commands of the coordinating agent. At the same time, it is connected to the vehicle model digital twin to transmit the actual chassis physical system status information to the chassis digital twin model in real time.

[0063] The digital twin of the driving environment is also connected to the perception module to acquire information related to the driving environment in real time. Based on the existing global map, it continuously updates and optimizes the twin model, which is used to build a virtual driving environment for the twin vehicle model.

[0064] The vehicle model digital twin is also connected to the chassis physical system to obtain real-time chassis status information and continuously update and optimize the twin model. Simultaneously, it is connected to the driving environment digital twin, enabling real-time information exchange between the two to accurately map the vehicle's driving state in the actual driving environment and obtain more precise vehicle status parameters.

[0065] In a comprehensive application example of the present invention, a multi-agent cooperative control method for a vehicle chassis cyber-physical system is provided, as shown in... Figure 4 The multi-agent coordinated control architecture shown is as follows: [The text abruptly ends here, so the translation stops as well.] Figure 5 As shown, the specific implementation steps are as follows:

[0066] 1) The perception module collects driving environment obstacle information, road information, real-time vehicle status information and driver command information and transmits them to the multi-agent coordination and control module, and transmits the driving environment information and road information containing obstacles to the driving environment twin model;

[0067] 2) The driving environment twin, based on the information received from the perception module and combined with the existing global map, establishes a precise local map of the current driving environment in real time, and constructs a virtual real-time driving environment.

[0068] 3) The multi-agent coordination control module receives information from the perception module and the vehicle model digital twin to change lane targets, with safety and comfort as constraints, to decide to call vehicle safety auxiliary systems, and to transmit the optimal cornering, optimal braking torque, and optimal suspension force to the drive-by-wire chassis physical system by coordinating the outputs of various vehicle safety auxiliary systems;

[0069] 4) The drive-by-wire chassis physical system receives control instructions from the multi-agent coordination control module, drives the drive-by-wire steering subsystem, the drive-by-wire braking subsystem, and the drive-by-wire suspension subsystem to execute the control instructions, and transmits chassis state information to the vehicle model digital twin;

[0070] 5) The vehicle model digital twin updates and optimizes the vehicle model digital twin in real time based on the received information from the drive-by-wire chassis physical system, and interacts with the driving environment digital twin in real time to truly map the driving state of the vehicle in the actual driving environment, and transmits the real-time running state information of the digital twin vehicle model to the multi-agent system control module;

[0071] The environmental obstacle information collected by the perception module in step 1) includes obstacle position and speed information, and road information includes lane information and road roughness information. Real-time state information of the ego vehicle mainly includes vehicle lateral, longitudinal, and vertical speed and acceleration, wheel speed and slip ratio, vehicle body yaw angular velocity, driving torque, braking torque, suspension damping force, suspension stiffness, suspension deflection, and suspension load. Environmental obstacle information is measured by laser radar and depth camera, and data fusion is performed in the perception module. Real-time state information of the ego vehicle and road information are measured by corresponding vehicle-mounted sensors, and state estimation and filtering processing are performed in the perception module.

[0072] The driving environment digital twin model in step 2) is placed in a cloud server, and information transmission is performed with the vehicle-mounted perception module through a 5G cellular network, the perception information is matched with the built-in global map, and a high-precision virtual driving environment is constructed.

[0073] Further, the multi-agent coordination control module in step 3) has the following specific working process:

[0074] 31) The perception module collects current driving environment information, real-time state information of the ego vehicle, and driver instruction information, and transmits them to the decision layer of the coordination control module. The decision layer first calculates the current acceleration of the ego vehicle, the relative distance and position between the ego vehicle and the obstacle, judges the current safety state of the ego vehicle, and predicts the safety state in a future preset time based on the vehicle state and driving environment information;

[0075] 32) If the vehicle state is determined to be stable and safe and the driver is in the driving mode, the multi-agent coordination control module directly transmits the information flow to the lower layer subsystem;

[0076] 33) If the vehicle state is determined to be stable and safe and the driver is in the driving mode, the multi-agent coordination control module directly transmits the information flow to the lower layer subsystem;

[0077] 34) If the vehicle state is determined to be unstable and unsafe, the decision layer selects to call the AEB system, the ESP system, the ABS system, the ASS system, the AFS system, the EPS system and the 4WS system in combination with the vehicle state and the driving environment information, to ensure the safety of the vehicle as the goal, to maintain the stability of the vehicle as the constraint, to coordinate the outputs of the intelligent agents of the vehicle safety auxiliary systems, to obtain the best steering angle, the best braking torque and the best suspension force, and to transmit to the lower layer subsystem.

[0078] In the process of coordinating the best control output of the intelligent agent, the observed current vehicle state is first input into the policy network, the policy network outputs the corresponding vehicle control action based on data training and obtains the corresponding reward, then the value network scores the control action and feeds back the scoring result as a supervision signal to the policy network to guide the policy network to update the network parameters, so that the action decision is continuously optimized. The vehicle control action output by the policy network is evaluated through the vehicle evaluation indexes based on safety, stability and comfort, and the evaluation indexes dynamically adjust their weights under different working conditions, as shown in the following formula.

[0079] J = aJ1 + bJ2 + gJ3

[0080] In the formula, J is the cost function, J1 is the safety evaluation index, J2 is the stability evaluation index, J3 is the comfort evaluation index, and a, b and g are the weight coefficients of each index.

[0081] When the vehicle executes a control command, the working condition environment will change, and the coordination control module executes the above control process again. When the cost function J reaches the minimum value, the output is the best control output.

[0082] The decision process of the best control output is as follows Figure 6As shown, after detecting the obstacle, first, the safety distance is judged, when it is the safety distance, if the driver does not operate, the car continues to run into the warning distance interval, at this time, the system warns the driver through the interactive interface signal light or steering wheel shaking, if the driver does not operate, the car continues to run into the dangerous distance interval, at this time, the multi-agent coordination control system works: the perception signal is transmitted to the multi-agent coordination control system, the system makes a decision based on the best slip rate, and further makes a comprehensive decision based on the stability judgment combined with suspension control and steering control and brake deceleration, and finally decides the best steering angle, the best suspension force and the best brake deceleration, which are executed by the brake motor, the steering motor and the suspension motor respectively.

[0083] The chassis state information in the step 4) is transmitted to the vehicle model digital twin located on the cloud server through the 5G cellular network, and the twin vehicle model is continuously corrected based on the real-time state of the real vehicle chassis;

[0084] The vehicle model digital twin in the step 5) is placed in the cloud server, and the twin vehicle model runs in the constructed twin driving environment, which is a real-time mapping of the vehicle running in the real driving environment.

[0085] Special application scenario one

[0086] The special application scenario one is a low adhesion road over-bend stability control working condition included in the comprehensive application example scenario, and the multi-agent coordination control method of the vehicle chassis information physical system of the application is used in the low adhesion road over-bend stability control working condition, and the multi-agent coordination control architecture is as shown in the figure Figure 7 The specific implementation steps are as follows:

[0087] 1) The perception module collects driving environment information, vehicle state information and driver instruction information and transmits them to the multi-agent coordination control module, and simultaneously transmits the driving environment information to the driving environment twin model;

[0088] 2) The driving environment twin establishes an accurate local map in the current driving environment in real time according to the information received from the perception module, and constructs a virtual real-time driving environment in combination with the existing global map. The constructed virtual driving environment includes accurate road information and traffic information such as road roughness, road curvature and lane line;

[0089] 3) The multi-agent coordination control module receives information from the perception module and the vehicle model digital twin, and makes a decision to call the AFS system and the ESP system based on the tracking and stability as indexes, coordinates the AFS agent and the ESP agent based on the multi-agent coordination control theory, comprehensively decides the best steering angle and the best brake torque, and transmits them to the line control chassis physical system.

[0090] 4) The line control chassis physical system receives control instructions from the multi-agent coordination control module, drives the line control braking subsystem and the line control steering subsystem to execute the control instructions, and transmits chassis state information to the vehicle model digital twin;

[0091] 5) The vehicle model digital twin updates and optimizes the vehicle model digital twin in real time according to the received information from the line control chassis physical system in combination with the existing virtual vehicle model, and the vehicle model digital twin interacts with the driving environment digital twin in real time, truly maps the driving state of the vehicle in the actual driving environment, and transmits the real-time running state information of the digital twin vehicle model to the multi-agent system control module.

[0092] Figure 8 The actual control effect of the multi-agent collaborative control method of the vehicle chassis information physical system based on the application is shown in the over-bend stability control working condition under low adhesion road surface. In the yaw angular velocity index representing the pros and cons of vehicle stability, the coordinated agent performs more stably than the AFS agent and the ESP agent, and in the lateral displacement index representing the path tracking performance, the error of the coordinated agent with the reference path is smaller than that of the AFS agent and the ESP agent.

[0093] Special application scenario two

[0094] The special application scenario two is a driver distraction human-machine co-driving control working condition included in the comprehensive application example scenario. The vehicle chassis information physical system architecture and the multi-agent collaborative control method of the application in the driver distraction human-machine co-driving control working condition have a multi-agent coordination control architecture as shown in Figure 9 The specific implementation steps are as follows:

[0095] 1) The perception module collects driving environment information, vehicle state information and driver instruction information and transmits them to the multi-agent coordination control module, and transmits the driving environment information to the driving environment twin model;

[0096] 2) The driving environment twin establishes an accurate local map in the current driving environment in real time according to the received information from the perception module in combination with the existing global map, and constructs a virtual real-time driving environment;

[0097] 3) The multi-agent coordination control module receives information from the perception module and the vehicle model digital twin, based on the driver's intention, and decides to select the 4WD system, the EPS system and the 4WS system based on the path tracking performance. Based on the multi-agent coordination control theory, the optimal steering angle and the optimal driving force are comprehensively decided by coordinating the 4WD system, the EPS agent and the 4WS agent, and are transmitted to the drive-by-wire chassis physical system;

[0098] 4) The drive-by-wire chassis physical system receives the control instructions from the multi-agent coordination control module, drives the drive-by-wire steering subsystem and the drive-by-wire driving subsystem to execute the control instructions, and transmits the chassis state information to the vehicle model digital twin;

[0099] 5) The vehicle model digital twin updates and optimizes the vehicle model digital twin in real time according to the information received from the drive-by-wire chassis physical system, and combines the existing virtual vehicle model. At the same time, the vehicle model digital twin interacts with the driving environment digital twin in real time, truly maps the driving state of the vehicle in the actual driving environment, and transmits the real-time running state information of the digital twin vehicle model to the multi-agent system control module.

[0100] Figure 10 The actual control effect of the multi-agent collaborative control method of the vehicle chassis information physical system based on the application is shown under the human-machine co-driving control working condition when the driver is distracted. In the lateral deviation and heading deviation indexes representing the path tracking performance, the deviation of the coordination agent is smaller than that of the EPS agent and the 4WS agent, and is more stable.

[0101] The application has many specific application ways, and the above description is only the preferred embodiment of the application. It should be pointed out that for ordinary skilled persons in the technical field, some improvements can be made without departing from the principle of the application, and these improvements should also be considered as the protection scope of the application.

[0102] It should be understood that although the present specification is described in terms of various embodiments, not every embodiment contains only one independent technical solution, and the description manner of the specification is only for the sake of clarity, and those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

[0103] The series of detailed descriptions listed above are only specific descriptions of the feasible embodiments of the application, and are not used to limit the protection scope of the application. Any equivalent embodiments or changes made without departing from the spirit of the application should be included in the protection scope of the application.

Claims

1. A multi-agent collaborative control method of a vehicle chassis cyber-physical system, characterized in that, The vehicle chassis information physical system comprises a perception module, a multi-agent coordinated control module, a drive-by-wire chassis physical system, a driving environment digital twin and a vehicle model digital twin. The perception module is connected with vehicle-mounted sensors and is used to collect vehicle state information and environment information; the perception module is also connected with the driving environment digital twin and is used to transmit real-time driving environment to the driving environment digital twin to build a virtual driving environment that is real-time synchronized with the real driving environment. The drive-by-wire chassis physical system is used to execute the control instructions of the multi-agent coordinated control module. The driving environment digital twin is provided with a global map, and the environment information is mapped into the driving environment digital twin in real time by the perception module to obtain a virtual real-time driving environment consistent with the real driving environment. The vehicle model digital twin is real-time interacted with the real drive-by-wire chassis physical system to real-map the driving state of the actual vehicle in the actual driving environment. The multi-agent coordinated control module selects and calls each automobile safety auxiliary system to control the drive-by-wire chassis physical system in combination with the current driving condition and driving environment. The multi-agent coordinated control module comprises an upper vehicle safety auxiliary system, a coordination agent and a decision layer; the coordination agent selects and calls each upper vehicle safety auxiliary system to control the vehicle according to the demand of the decision layer. The method comprises the following steps: 1) The perception module collects driving environment information and road information and transmits them to the driving environment twin model, collects vehicle state information and driver instruction information and transmits them to the multi-agent coordinated control module; 2) The driving environment twin establishes a local map under the current driving environment in real time according to the information received from the perception module in combination with the existing global map to build a virtual real-time driving environment; 3) The multi-agent coordinated control module receives the information from the perception module and the information from the vehicle model digital twin, decides to select and call the vehicle safety auxiliary system, and transmits the best output control quantity to the drive-by-wire chassis physical system through the coordination of the output of each vehicle safety auxiliary system by the coordination agent; 4) The drive-by-wire chassis physical system receives the control instructions from the multi-agent coordinated control module, executes the control instructions through the driving actuator, and transmits the chassis state information to the vehicle model digital twin; 5) The vehicle model digital twin updates and optimizes the vehicle model digital twin in real time according to the information received from the drive-by-wire chassis physical system in combination with the existing virtual vehicle model, and the vehicle model digital twin is real-time interacted with the driving environment digital twin to real-map the driving state of the vehicle in the actual driving environment, and transmits the real-time running state information of the twin vehicle model to the multi-agent coordinated control module; The specific working process of the multi-agent coordinated control module in step 3) comprises: 31) The perception module collects current driving environment information, vehicle state information and driver instruction information, which are transmitted to the decision layer of the multi-agent coordination control module. The decision layer first combines vehicle state and driving environment information to calculate the current acceleration of the vehicle and the relative distance and position between the vehicle and the obstacle, and then judges the stability of the current vehicle state and predicts the stability of the vehicle state in a future preset time; 32) If the vehicle state is determined to remain stable and the driver is in driving mode, the multi-agent coordination control module directly transmits the information stream to the physical system of the drive-by-wire chassis; 33) If the vehicle state is determined to remain stable and the driver is in automatic driving mode, the decision layer selects one or more vehicle safety auxiliary systems that maintain the stable and safe driving of the vehicle, and the coordination agent coordinates the output of each vehicle safety auxiliary system; 34) If the vehicle state is determined to be unstable, the decision layer selects one or more vehicle safety auxiliary systems that maintain the stability of the vehicle state, and the coordination agent coordinates the output of each vehicle safety auxiliary system.

2. The multi-agent collaborative control method of vehicle chassis cyber-physical system according to claim 1, characterized in that, The perception module information in step 1) includes external environment information, vehicle state information and driver control instruction information; The external environment information is measured by laser radar and depth camera, and the data is fused in the perception module; the vehicle state information is measured by the vehicle-mounted sensor, and the state estimation and filtering processing are performed in the perception module.

3. The multi-agent collaborative control method for vehicle chassis cyber-physical system according to claim 1, wherein, The driving environment twin model in step 2) is set in the cloud server, and the perception information is matched with the built-in global map through 5G cellular network to transmit information to the vehicle-mounted perception module, and the virtual driving environment is constructed.

4. The multi-agent collaborative control method of vehicle chassis cyber-physical system according to claim 1, wherein, The chassis state information in step 4) is transmitted to the vehicle model digital twin located on the cloud server through the 5G cellular network, and the twin vehicle model is continuously corrected based on the real-time state of the real vehicle chassis.

5. The multi-agent collaborative control method of vehicle chassis cyber-physical system according to claim 1, wherein, The vehicle model digital twin in step 5) is placed in the cloud server, and the twin vehicle model runs in the constructed twin driving environment, which is a real-time mapping of the vehicle running in the real driving environment.

6. The multi-agent collaborative control method of vehicle chassis cyber-physical system according to claim 1, wherein, The multi-agent coordination control module includes upper vehicle safety auxiliary systems, coordination agents and a decision layer; the coordination agent selects and coordinates each upper vehicle safety auxiliary system according to the demand of the decision layer to control the vehicle. 7.The multi-agent collaborative control method of vehicle chassis cyber-physical system according to claim 1, wherein, The drive-by-wire chassis physical system includes a drive-by-wire steering subsystem, a wheel system, a drive-by-wire braking subsystem, a drive-by-wire suspension subsystem and a drive-by-wire driving subsystem; the vehicle model digital twin includes a drive-by-wire driving twin, a drive-by-wire suspension twin, a drive-by-wire braking twin, a drive-by-wire steering twin and four wheel twins. 8.The multi-agent collaborative control method of vehicle chassis cyber-physical system according to claim 1, wherein, The automobile safety auxiliary system includes an emergency braking system, a vehicle body stability system, an anti-lock braking system, a front wheel active steering system, a four-wheel steering system, an intelligent automobile comprehensive auxiliary steering system, a four-wheel drive system, an adaptive cruise system, an active suspension system and a driving force control system.

Images