Intelligent chassis drift collision avoidance control method and system and electronic and electrical architecture

Abstract

An intelligent chassis drift collision avoidance control method and system and an electronic and electrical architecture. The method comprises: performing quantitative analysis on collision avoidance requirement information by means of an obstacle envelope method so as to obtain obstacle envelope data; calculating a normal operation infeasible state domain of the obstacle envelope data on the basis of a vehicle dynamics model; performing safety analysis on the basis of the normal operation infeasible state domain, so as to determine, on the basis of a safety analysis result, whether to start an extreme collision avoidance operation; and if it is determined that the extreme collision avoidance operation is started, performing autonomous decision-making control of a vehicle on the basis of extreme dynamics control input data generated by a chassis domain control platform, so as to achieve drift collision avoidance. The method can give full play to the collision avoidance potential of vehicles, and improve the emergency response capability and backup level of intelligent vehicles.

Description

An intelligent chassis drift collision avoidance control method, system and electronic and electrical architecture

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This disclosure is based on and claims the priority of Chinese patent application with application number 202410309199.X and application date March 18, 2024. The entire content of the Chinese patent application is hereby incorporated into this disclosure as a reference. Technical Field

[0003] The present disclosure relates to the field of active automobile safety technology, and in particular to an intelligent chassis drift and collision avoidance control method, system, and electronic and electrical architecture. Background Art

[0004] According to the World Health Organization, road traffic injuries are the leading cause of death among young people aged 5 to 29, resulting in approximately 1.3 million deaths annually. Of these, driver error is responsible for up to 93% of traffic accidents. Autonomous driving technology holds great promise for reducing or preventing accidents caused by human error, thereby improving vehicle safety. However, even autonomous vehicles equipped with advanced perception systems and safe driving protocols may face situations requiring split-second decision-making and action in complex high-speed road conditions, exceeding the capabilities of traditional active safety features.

[0005] To address these challenges, a variety of sophisticated active safety features have been developed, such as anti-lock braking systems, automatic emergency braking systems, and electronic stability systems, designed to prevent instability and collisions in critical situations. Active safety features are crucial during emergency collision avoidance. The primary purpose of these active safety mechanisms is to significantly reduce vehicle speed or stabilize the vehicle.

[0006] However, these functions primarily restrict the vehicle to a linear state of lateral stability and fail to fully exploit the vehicle's full maneuverability potential. In some particularly critical situations, maneuvers beyond this stable state may be necessary. When necessary, expert drivers can perform collision avoidance maneuvers through aggressive steering, resulting in a degree of vehicle sideways slip, commonly referred to as "drifting." Vehicle limit control is designed to improve vehicle maneuverability in high-sideslip conditions through aggressive steering, thrusting, and braking. Such extreme maneuvers pose a significant challenge to the operating system and may result in a poor passenger experience. Therefore, in autonomous buses, drifting should only be used when absolutely necessary.

[0007] Currently, theoretical approaches for collision avoidance control using extreme maneuvers have yet to be established, and the triggering criteria, control methods, and exit mechanisms for these maneuvers remain unclear. Regarding triggering criteria, some studies use reachability to analyze vehicle safety, but this simplifies the vehicle into a point-mass model, making it difficult to quantitatively analyze extreme maneuvers. Regarding control methods, existing research mostly uses model-based or data-driven control strategies, but their online deployment faces challenges or skepticism. Regarding exit mechanisms, existing research uses parking as the final state and does not consider integration with other autonomous driving systems.

[0008] Summary of the Invention

[0009] The present disclosure aims to solve one of the technical problems in the related art at least to a certain extent.

[0010] To this end, the present disclosure proposes an intelligent chassis drift collision avoidance control method to fully tap the vehicle's collision avoidance potential and enhance the emergency response capability and backup level of the intelligent vehicle.

[0011] Another object of the present disclosure is to provide an intelligent chassis drift and collision avoidance control system.

[0012] The third objective of the present disclosure is to provide an electronic and electrical architecture.

[0013] To achieve the above objectives, the present disclosure provides, on one hand, an intelligent chassis drift collision avoidance control method, comprising:

[0014] Obstacle envelope method is used to quantitatively analyze collision avoidance requirement information to obtain obstacle envelope data;

[0015] Calculating a conventional operation infeasible state domain of the obstacle envelope data based on a vehicle dynamics model;

[0016] Performing a safety analysis based on the conventional operation infeasible state domain to determine whether to initiate an extreme collision avoidance maneuver according to the safety analysis result;

[0017] If it is determined that an extreme collision avoidance operation is to be initiated, the vehicle will be autonomously decided and controlled based on the extreme dynamics control input data generated by the chassis domain control platform to achieve drift collision avoidance.

[0018] The intelligent chassis drift collision avoidance control method of the disclosed embodiment may also have the following additional technical features:

[0019] In one embodiment of the present disclosure, the obstacle envelope method is used to quantitatively analyze the collision avoidance requirement information to obtain obstacle envelope data, including:

[0020] Identify the location and shape of obstacles in the vehicle's surrounding environment based on the fusion results of the vehicle's lidar raw data and the Internet of Vehicles data;

[0021] Determining a minimum circular envelope that completely encloses the detected obstacles based on the obstacle position and obstacle shape;

[0022] The minimum circular envelope is identified to obtain obstacle envelope data; wherein the obstacle envelope data includes the center coordinates and radius of the envelope.

[0023] In one embodiment of the present disclosure, calculating the conventional operation infeasible state domain of the obstacle envelope data based on the vehicle dynamics model includes:

[0024] establishing a vehicle dynamics model, and determining vehicle motion data based on the vehicle dynamics model;

[0025] generating restriction conditions for vehicle status and action of conventional operations based on the vehicle motion data and preset road surface data;

[0026] An optimal control problem is constructed based on the constraint conditions, the obstacle envelope data, and the current vehicle speed, and a numerical method is used to solve the optimal control problem to generate an infeasible state domain for conventional operation.

[0027] In one embodiment of the present disclosure, autonomous decision-making control of a vehicle to achieve drift collision avoidance based on extreme dynamics control input data generated by a chassis domain control platform includes:

[0028] The extreme dynamics control input data generated by the chassis domain control platform is added to the conventional control term as a residual term, and the strength of the residual term is used to judge the degree of intervention of the extreme dynamics control;

[0029] Using vehicle data under high side deviation, calibrate multiple sets of model parameters under high side deviation and normal conditions, and set switching rules; and

[0030] Using rule-based, model-based or data-driven autonomous decision-making control algorithms, residual terms are generated after the extreme dynamics are triggered. The corresponding optimization objectives are the collision avoidance performance index and the degree of intervention index on the original driving goal.

[0031] In one embodiment of the present disclosure, the collision avoidance performance index includes the degree of proximity to the boundary of the infeasible area of ​​the obstacle envelope; the intervention degree index with respect to the original driving goal includes the degree of intervention of the extreme dynamics control and the difficulty of the automatic driving system taking over after exiting the intervention.

[0032] To achieve the above objectives, the present disclosure further provides an intelligent chassis drift and collision avoidance control system, comprising:

[0033] an obstacle data determination module, configured to quantitatively analyze the collision avoidance requirement information using an obstacle envelope method to obtain obstacle envelope data;

[0034] an infeasible state domain calculation module, configured to calculate an infeasible state domain of a conventional operation of the obstacle envelope data based on a vehicle dynamics model;

[0035] A safety analysis and judgment module is used to perform a safety analysis based on the conventional operation infeasible state domain, so as to determine whether to initiate an extreme collision avoidance operation according to the safety analysis result;

[0036] The autonomous decision-making control module is used to determine whether to initiate extreme collision avoidance operations. It performs autonomous decision-making control of the vehicle based on the extreme dynamics control input data generated by the chassis domain control platform to achieve drift collision avoidance.

[0037] To achieve the above objectives, the third aspect of the present disclosure proposes an electronic and electrical architecture, including: a chassis domain control platform, an on-board computing platform, an Internet of Vehicles control platform, a vehicle execution system, a vehicle sensing system, and a Gigabit Ethernet on-board transmission system;

[0038] The chassis domain control platform is used to coordinate and control the vehicle's power system, braking system, and suspension system, as well as autonomous decision-making and control under extreme working conditions;

[0039] The vehicle-mounted computing platform is used to process raw data from various sensors and cameras;

[0040] The Internet of Vehicles control platform is used for external communication;

[0041] The vehicle sensing system is used to transmit raw data to the chassis domain control platform via the Gigabit Ethernet vehicle transmission system;

[0042] The chassis domain control platform analyzes and responds to the raw data, shields other active safety functions and stability functions of the vehicle, and sends collision avoidance action instructions to the vehicle execution system;

[0043] The vehicle execution system is used to perform drift collision avoidance in response to the collision avoidance action instruction.

[0044] The intelligent chassis drift collision avoidance control method, system, and electronic and electrical architecture of the disclosed embodiments achieve autonomous decision-making control for extreme collision avoidance through advanced electronic and electrical architecture and sophisticated models. While ensuring generalization, it minimizes the impact on the original driving task and provides highly accurate and rapid response.

[0045] Additional aspects and advantages of the present disclosure will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The above and / or additional aspects and advantages of the present disclosure will become apparent and readily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0047] FIG1 is a flow chart of an intelligent chassis drift collision avoidance control method according to an embodiment of the present disclosure;

[0048] FIG2 is a structural diagram of an intelligent chassis drift and collision avoidance control system according to an embodiment of the present disclosure;

[0049] FIG3 is a structural diagram of another intelligent chassis drift and collision avoidance control system according to an embodiment of the present disclosure;

[0050] FIG4 is a diagram of an automotive electronic and electrical architecture for intelligent chassis drift and collision avoidance control according to an embodiment of the present disclosure;

[0051] FIG5 is a diagram showing the collision avoidance effect according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0052] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in the present disclosure can be combined with each other. The present disclosure will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0053] In order to enable those skilled in the art to better understand the solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the embodiments described are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present disclosure.

[0054] The following describes an intelligent chassis drift collision avoidance control method, system, and electronic and electrical architecture proposed according to embodiments of the present disclosure with reference to the accompanying drawings.

[0055] FIG1 is a flow chart of an intelligent chassis drift collision avoidance control method according to an embodiment of the present disclosure.

[0056] As shown in FIG1 , the method includes but is not limited to the following steps:

[0057] S1, using the obstacle envelope method to quantitatively analyze the collision avoidance requirement information to obtain obstacle envelope data;

[0058] S2, the infeasible state domain of conventional operation based on the calculation of obstacle envelope data by the vehicle dynamics model;

[0059] S3, performing safety analysis based on the infeasible state domain of conventional operations to determine whether to initiate extreme collision avoidance operations based on the safety analysis results;

[0060] S4, judging to start the extreme collision avoidance operation, then performing autonomous decision-making control of the vehicle based on the extreme dynamics control input data generated by the chassis domain control platform to achieve drift collision avoidance.

[0061] Specifically, the intelligent chassis drift collision avoidance control method disclosed in the present invention quantifies the collision avoidance requirements and calculates the infeasible state domain based on a fine model as the quantification basis for triggering extreme driving; the accessibility-guided extreme collision avoidance autonomous decision-making control ensures minimal impact on the original driving task while expanding generalization, and monitors the controllability of the vehicle when performing extreme autonomous decision-making control based on the monitoring of the extreme dynamic safety boundary.

[0062] In one embodiment of the present disclosure, the collision avoidance requirements are quantified, including: first, fusing the raw lidar data and the Internet of Vehicles data to quickly and in real time identify the location and shape of obstacles in the surrounding environment; generating a minimum circular envelope that is large enough to enclose all detected obstacles; identifying and recording the center coordinates and radius of the envelope; this information is crucial for calculating the collision avoidance path and making emergency maneuvering decisions.

[0063] In one embodiment of the present disclosure, a vehicle dynamics model is established based on a refined model to calculate an infeasible state domain as a quantified basis for triggering extreme driving. This domain includes, but is not limited to, yaw, roll, pitch, lateral motion, and longitudinal motion. Road conditions are combined to generate constraints on vehicle state and motion imposed by conventional maneuvers. An optimal control problem is constructed based on the location and radius of the obstacle envelope, conventional maneuvering constraints, and the current vehicle speed. Numerical methods are used to solve the optimal control problem and generate an infeasible state domain for conventional maneuvers. Furthermore, a reachable value function is generated as a reference for subsequent decision-making and control.

[0064] Furthermore, the generated boundary of the infeasible state region for conventional operation is used as the theoretical basis for extreme dynamics control. Entering the infeasible state region is the condition for triggering extreme dynamics control, and exiting the infeasible state region is the condition for exiting extreme dynamics control.

[0065] In one embodiment of the present disclosure, the extreme collision avoidance autonomous decision control, while expanding generalization, ensures minimal impact on the original driving task, which includes: residual control, adding the extreme dynamics control input as a residual term to the conventional control term, and then using the strength of the residual term as a direct indicator to judge the degree of intervention of the extreme dynamics control; using offline vehicle data to calibrate or correct model accuracy: collecting vehicle data under high side deviation, using this data set to calibrate multiple sets of parameters of the model under high side deviation and normal conditions, and setting switching rules; optimal autonomous decision control: using rule-based, model-based or data-driven autonomous decision control algorithms, responsible for generating residual terms after the extreme dynamics are triggered, and its optimization objectives include collision avoidance performance and the degree of intervention in the original driving goal. In addition, the reachable value information is used as guidance information for decision control to accelerate the solution process or improve the generalization of the controller. e =u n +a t

[0066] The residual control is as described above, where u n is the control input for conventional autonomous driving, a t is the residual input, a e is the final control input.

[0067] This reachability information serves as a reference for decision-making and control, including but not limited to using it to initialize a reinforcement learning critic network, integrating it into the optimal control objective function, or using it as a logical judgment item in rules. This reachability-guided strategy will increase the difficulty of solving autonomous decision-making and control, and allocate more computing resources to ensure generalization.

[0068] The collision avoidance performance metric primarily includes the distance from the obstacle envelope and the proximity to the boundary of the infeasible region. Specifically, a higher distance from the obstacle envelope is rewarded, indicating a lower collision risk; a lower proximity to the boundary of the infeasible region is rewarded, indicating a closer approach to collision avoidance.

[0069] The degree of intervention in the original driving objective primarily includes the degree of involvement of extreme dynamics control and the difficulty of the autonomous driving system taking over after exiting the intervention. Specifically, while prioritizing collision avoidance, a lower degree of extreme dynamics control intervention is rewarded. A smaller angle between the vehicle's yaw angle and the tangent of the obstacle envelope when exiting extreme dynamics control is rewarded, indicating a lower takeover difficulty.

[0070] Furthermore, the chassis domain control platform generates extreme dynamic control inputs, shielding all other active safety functions of the vehicle, including but not limited to automatic emergency braking, anti-lock braking system and electronic stability system; and sends input signals to the chassis's execution systems: drive system, braking system, steering system and suspension system.

[0071] Furthermore, the system monitors and constrains extreme dynamic safety margins. Specifically, this includes a predefined dynamic safety margin table, primarily based on the vehicle's yaw rate and sideslip angle. If the vehicle's dynamic state exceeds the corresponding margin, high-priority stability control is activated until the vehicle returns to the safety margin.

[0072] According to the intelligent chassis drift collision avoidance control method of the embodiment of the present disclosure, the vehicle safety in extreme and emergency situations is significantly improved by introducing intelligent chassis drift collision avoidance control. It can effectively deal with complex road conditions that are beyond the processing range of traditional algorithms, thereby reducing or avoiding accidents caused by driver errors. The method of quantifying the infeasible area of ​​conventional operations can provide a theoretical basis for triggering extreme operations, ensuring that extreme operations are enabled in necessary scenarios. Through advanced electronic and electrical architecture and sophisticated models, autonomous decision-making control of extreme collision avoidance is achieved, while ensuring generalization, minimizing the impact on the original driving task, and providing highly accurate and fast response. The present disclosure comprehensively utilizes lidar raw data, vehicle network data and advanced computing models to achieve rapid and real-time recognition of the surrounding environment. This data integration and processing capability enables the system to effectively evaluate and respond to sudden road conditions and obstacles, enhancing the emergency response capability and backup level of autonomous driving vehicles.

[0073] To implement the above embodiment, as shown in FIG2 , this embodiment further provides an intelligent chassis drift collision avoidance control system 10, which includes:

[0074] The obstacle data determination module 100 is configured to quantitatively analyze the collision avoidance requirement information using an obstacle envelope method to obtain obstacle envelope data;

[0075] An infeasible state domain calculation module 200 is used to calculate the conventional operation infeasible state domain of the obstacle envelope data based on the vehicle dynamics model;

[0076] A safety analysis and judgment module 300 is used to perform a safety analysis based on the conventional operation infeasible state domain, and to determine whether to initiate an extreme collision avoidance maneuver based on the safety analysis results;

[0077] The autonomous decision control module 400 is used to determine whether to start an extreme collision avoidance operation, and then perform autonomous decision control of the vehicle based on the extreme dynamics control input data generated by the chassis domain control platform.

[0078] Furthermore, the obstacle data determination module 100 is further configured to:

[0079] Identify the location and shape of obstacles in the vehicle's surrounding environment based on the fusion results of the vehicle's lidar raw data and the Internet of Vehicles data;

[0080] Determining a minimum circular envelope that completely encloses the detected obstacles based on the obstacle position and obstacle shape;

[0081] The minimum circular envelope is identified to obtain obstacle envelope data; wherein the obstacle envelope data includes the center coordinates and radius of the envelope.

[0082] Furthermore, the infeasible state domain calculation module 200 is further configured to:

[0083] establishing a vehicle dynamics model, and determining vehicle motion data based on the vehicle dynamics model;

[0084] generating restriction conditions for vehicle status and action of conventional operations based on the vehicle motion data and preset road surface data;

[0085] An optimal control problem is constructed based on the constraint conditions, the obstacle envelope data, and the current vehicle speed, and a numerical method is used to solve the optimal control problem to generate an infeasible state domain for conventional operation.

[0086] Furthermore, the autonomous decision-making control module 400 is further configured to:

[0087] The extreme dynamics control input data generated by the chassis domain control platform is added to the conventional control term as a residual term, and the degree of intervention of the extreme dynamics control is judged based on the strength of the residual term.

[0088] Using vehicle data under high side deviation, calibrate multiple sets of model parameters under high side deviation and normal conditions, and set switching rules; and

[0089] Using rule-based, model-based or data-driven autonomous decision-making control algorithms, residual terms are generated after the extreme dynamics are triggered. The corresponding optimization objectives are the collision avoidance performance index and the degree of intervention index on the original driving goal.

[0090] FIG3 is an architecture diagram of another intelligent chassis drift collision avoidance control system according to an embodiment of the present disclosure, as shown in FIG3 :

[0091] Module 21 quantifies the collision avoidance requirements in the environment and uses the obstacle envelope method for quantitative analysis;

[0092] Module 22 calculates the conventional operation infeasible state domain of the obstacle envelope and uses the state domain to perform safety analysis to determine whether to initiate extreme collision avoidance maneuvers;

[0093] After the extreme collision avoidance maneuver is triggered, the extreme maneuver controller 23 performs autonomous decision-making and control;

[0094] During the execution of extreme maneuvers, active safety functions such as the conventional collision avoidance system 24 and the stability limiting function 25 are all shielded and monitored and constrained by the dynamic safety boundary determination module 26 .

[0095] According to the intelligent chassis drift collision avoidance control system of the embodiment of the present disclosure, the vehicle safety in extreme and emergency situations is significantly improved by introducing intelligent chassis drift collision avoidance control. It can effectively deal with complex road conditions that are beyond the processing range of traditional algorithms, thereby reducing or avoiding accidents caused by driver errors. The method of quantifying the infeasible area of ​​conventional operations can provide a theoretical basis for the triggering of extreme operations, ensuring that extreme operations are enabled in necessary scenarios. Through advanced electronic and electrical architecture and sophisticated models, autonomous decision-making control of extreme collision avoidance is achieved, while ensuring generalization, minimizing the impact on the original driving task, and providing highly accurate and fast response. The present disclosure comprehensively utilizes lidar raw data, vehicle network data and advanced computing models to achieve rapid and real-time recognition of the surrounding environment. This data integration and processing capability enables the system to effectively evaluate and respond to sudden road conditions and obstacles, enhancing the emergency response capability and backup level of autonomous driving vehicles.

[0096] In one embodiment of the present disclosure, the electronic and electrical architecture of the intelligent chassis drift and collision avoidance control of the present disclosure may include: a chassis domain control platform, an on-board computing platform, an Internet of Vehicles control platform, a vehicle execution system, a vehicle sensing system, and a Gigabit Ethernet on-board transmission system;

[0097] The chassis domain control platform is used to coordinate and control the vehicle's power system, braking system, and suspension system, as well as autonomous decision-making and control under extreme working conditions;

[0098] The vehicle-mounted computing platform is used to process raw data from various sensors and cameras;

[0099] The Internet of Vehicles control platform is used for external communication;

[0100] The vehicle sensing system is used to transmit raw data to the chassis domain control platform via the Gigabit Ethernet vehicle transmission system;

[0101] The chassis domain control platform analyzes and responds to the raw data, shields other active safety functions and stability functions of the vehicle, and sends collision avoidance action instructions to the vehicle execution system;

[0102] The vehicle execution system is used to perform drifting to avoid collision in response to a collision avoidance action instruction.

[0103] Specifically, FIG4 is an automotive electronic and electrical architecture adapted to extreme drift collision avoidance control of an autonomous vehicle according to an embodiment of the present disclosure, as shown in FIG4 :

[0104] This system architecture includes ultrasonic radar, millimeter-wave radar, lidar, cameras, an onboard intelligent computing platform, and a connected vehicle control platform. After extreme dynamics control is triggered, the signal flow is shown in orange. Raw data from the lidar and vehicle network is transmitted via Gigabit Ethernet to the chassis domain controller. The chassis domain controller calculates throttle position, brake pressure, and front wheel angle, and transmits this data to the execution system.

[0105] In this embodiment, the chassis domain controller analyzes the raw data of the perception system and determines that the obstacle envelope center is (X0, Y0) and the obstacle envelope radius is R. The obstacle envelope Γ can be expressed as:

[0106] In this embodiment, the infeasible state domain of the obstacle envelope Γ is recorded as but The mathematical expression is:

[0107] Where x represents the initial state of the vehicle, and γ(·) represents the disturbance encountered during driving. The set ζ contains the time from t = 0 to t = t f , the states along the vehicle trajectory during the period, which are determined by the vehicle dynamics system control, in response to the control sequence u(·) and the disturbance γ(·).

[0108] In this embodiment, the infeasible state domain corresponding to the obstacle envelope is obtained by discretization and pre-traversal. And store it in storage for real-time query.

[0109] In this embodiment, a method for autonomous decision-making and control of extreme collision avoidance employs a reinforcement learning controller. It is worth noting that the autonomous decision-making and control method for extreme collision avoidance protected by this disclosure also includes rule-based and model-based controllers. Controls based on reachability and related control architectures are also within the scope of this disclosure.

[0110] The control strategy is designed based on reinforcement learning. The state space S includes all the information required for the intelligent chassis to drift and avoid collisions, which is obtained by the chassis domain controller analyzing the sensor system data. The state space of this embodiment is as follows: S = [s ego ,s enr ] s ego =[v x ,v y ,r,γ,X e ,Y e ,ψ,φ,s,u n ] s enr =[d r ,d b ]

[0111] Where s ego ,s enr u is the vehicle state information and collision avoidance information required in this embodiment; n is the output of the conventional autonomous driving system; v x ,v y , r, γ are the longitudinal velocity, lateral velocity, yaw angular velocity, and roll velocity of the vehicle in the vehicle coordinate system respectively; X e ,Y e ,ψ,φ are the longitudinal position, lateral position, yaw angle and roll angle of the ego vehicle in the world coordinate system respectively; s is the state of the failed vehicle, including: collision avoidance failure, collision avoidance success, collision avoidance process, and rollover; d r is the shortest distance between the vehicle and the obstacle; d b is the distance between the vehicle and the boundary of the infeasible state domain. r It can reflect the danger level of the vehicle’s current state, d b Reflects how close the vehicle is to successfully avoiding a collision.

[0112] The vehicle studied in this embodiment is a four-wheel steering vehicle, and the action space A is constructed as follows: A=[δ f ,δ r ,T f ,T r ]

[0113] Where, δ f ,δ r is the steering angle of the front and rear wheels; T f ,T r is the driving / braking torque of the front and rear wheels.

[0114] This embodiment adopts the residual input method, and the final input a e The calculation method is: e =u n +a t

[0115] Among them, a t is the residual input output by reinforcement learning, u n is the output of the conventional autonomous driving system. t The absolute value of the item can be directly used as a quantitative indicator to judge the degree of intervention of extreme operations.

[0116] A high-precision eight-degree-of-freedom brake failure vehicle dynamics model is constructed as the simulation environment. The vehicle body model is a dual-track four-degree-of-freedom vehicle body model:

[0117] Where, F x1 ,F x2 ,Fx3 ,F x4 The longitudinal force provided to the four-wheel tires; F y1 ,F y2 ,F y3 ,F y4 The lateral force provided to the four-wheel tires; L a ,L b are the distances from the front axle to the center of mass and the distance from the rear axle to the center of mass respectively; W f ,W r are the front wheel spacing and rear wheel spacing respectively; h is the height of the vehicle's center of mass. M x =mghφ+(-D f γ-K f φ)+(-D r γ-K r φ)

[0118] Where ρ is the air density, C d is the air resistance coefficient, A is the cross-sectional area of ​​the vehicle; D f ,D r are forward roll damping and backward roll damping respectively; K f ,K r are the forward rolling stiffness and the backward rolling stiffness respectively.

[0119] The tire model uses a magic formula fitted with experimental data, which will not be described in detail.

[0120] Reward function design: The reward function setting includes the final reward R T and instant reward R i R=R i +R T

[0121] Instant rewards include: collision avoidance item R i1 , used to reward staying away from obstacles; collision avoidance item R i2 , used to reward the vehicle's successful approach to collision avoidance and reward approaching the boundary of the infeasible state domain; the intervention degree term R i3 , used to punish excessive intervention; tire friction utilization term R i4 , used to indicate the degree of closeness between the vehicle's motion state and the current limit braking performance. Specific descriptions are as follows:

[0122] Collision avoidance term R i1 , rewards the vehicle with a larger distance from the obstacle in the world coordinate system. The larger the distance, the higher the vehicle's immediate safety. i1 =k1·d r

[0123] Collision avoidance term R i2, rewards vehicles that are close to the boundary of the infeasible state domain. The smaller the distance, the closer the current vehicle state is to successful collision avoidance. R i2 =k2·d b

[0124] Intervention level item R i3 , punishing the excessive degree of intervention of reinforcement learning control items; R i3 =k3·a t

[0125] Tire friction utilization term R i4 , rewards the degree to which the vehicle's motion state is close to the current limit braking performance.

[0126] Final rewards include: completing the collision avoidance bonus item R T1 , takeover difficulty penalty item R T2 .

[0127] Where k5 is a positive number, and a high-level reward is given if no collision occurs; k6 and k7 are negative numbers, and a high-level penalty is given if rollover, collision, or exceeding the road boundary occurs.

[0128] ψ is the current yaw angle of the car. This penalty encourages the vehicle to have a small angle between the yaw angle and the tangent of the obstacle when driving out of the infeasible area, which means that the autonomous driving system has low takeover difficulty.

[0129] Build the Critic and Actor deep neural networks for the reinforcement learning algorithm. Determine the reward R based on the state S and action A. Train the Critic network parameters by minimizing the temporal difference loss function; and train the Actor network parameters by maximizing the value function.

[0130] In a real-world environment, the trained network parameters are imported into the on-board computer. After drift control is triggered, the chassis domain controller blocks the vehicle's other active safety and stability functions and sends collision avoidance instructions to the vehicle actuators to achieve drift collision avoidance.

[0131] During drift avoidance, the dynamic safety boundary provides stability constraints. This embodiment ensures this by limiting the vehicle's sideslip angle and yaw rate to within the open-loop stability boundary. Figure 5 illustrates the control effects of this embodiment in a complex and critical collision avoidance scenario.

[0132] According to the electronic and electrical architecture of the intelligent chassis drift collision avoidance control of the embodiment of the present disclosure, the vehicle safety in extreme and emergency situations is significantly improved by introducing intelligent chassis drift collision avoidance control. It can effectively deal with complex road conditions that are beyond the processing range of traditional algorithms, thereby reducing or avoiding accidents caused by driver errors. The method of quantifying the infeasible area of ​​conventional operations can provide a theoretical basis for triggering extreme operations, ensuring that extreme operations are enabled in necessary scenarios. Through advanced electronic and electrical architecture and sophisticated models, autonomous decision-making control of extreme collision avoidance is achieved, while ensuring generalization, minimizing the impact on the original driving task, and providing highly accurate and fast response. The present disclosure comprehensively utilizes lidar raw data, vehicle network data and advanced computing models to achieve rapid and real-time recognition of the surrounding environment. This data integration and processing capability enables the system to effectively evaluate and respond to sudden road conditions and obstacles, enhancing the emergency response capability and backup level of autonomous driving vehicles.

[0133] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification and features of different embodiments or examples, unless they are mutually inconsistent.

[0134] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. Throughout the present disclosure, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

Claims

1. An intelligent chassis drift collision avoidance control method, comprising: Obstacle envelope method is used to quantitatively analyze collision avoidance requirement information to obtain obstacle envelope data; Calculating a conventional operation infeasible state domain of the obstacle envelope data based on a vehicle dynamics model; Performing a safety analysis based on the conventional operation infeasible state domain to determine whether to initiate an extreme collision avoidance maneuver according to the safety analysis result; If it is determined that an extreme collision avoidance operation is to be initiated, the vehicle will be autonomously decided and controlled based on the extreme dynamics control input data generated by the chassis domain control platform to achieve drift collision avoidance.

2. The method according to claim 1, wherein the obstacle envelope method is used to quantitatively analyze the collision avoidance requirement information to obtain obstacle envelope data, comprising: Identify the location and shape of obstacles in the vehicle's surrounding environment based on the fusion results of the vehicle's lidar raw data and the Internet of Vehicles data; Determining a minimum circular envelope that completely encloses the detected obstacles based on the obstacle position and obstacle shape; The minimum circular envelope is identified to obtain obstacle envelope data; wherein the obstacle envelope data includes the center coordinates and radius of the envelope.

3. The method according to claim 1 or 2, wherein calculating the conventional operation infeasible state domain of the obstacle envelope data based on a vehicle dynamics model comprises: establishing a vehicle dynamics model, and determining vehicle motion data based on the vehicle dynamics model; generating restriction conditions for vehicle status and action of conventional operations based on the vehicle motion data and preset road surface data; An optimal control problem is constructed based on the constraint conditions, the obstacle envelope data, and the current vehicle speed, and a numerical method is used to solve the optimal control problem to generate an infeasible state domain for conventional operation.

4. The method according to any one of claims 1 to 3, wherein performing autonomous decision-making control of the vehicle to achieve drift collision avoidance based on extreme dynamics control input data generated by a chassis domain control platform comprises: The extreme dynamics control input data generated by the chassis domain control platform is added to the conventional control term as a residual term, and the strength of the residual term is used to judge the degree of intervention of the extreme dynamics control; Using vehicle data under high side deviation, calibrate multiple sets of model parameters under high side deviation and normal conditions, and set switching rules; and Using rule-based, model-based or data-driven autonomous decision-making control algorithms, residual terms are generated after the extreme dynamics are triggered. The corresponding optimization objectives are the collision avoidance performance index and the degree of intervention index on the original driving goal.

5. The method according to claim 4, wherein the collision avoidance performance index includes the proximity of the distance to the obstacle envelope to the boundary of the infeasible region; the intervention degree index for the original driving goal includes the degree of intervention of the extreme dynamics control and the difficulty of the autonomous driving system taking over after exiting the intervention.

6. An intelligent chassis drift and collision avoidance control system, comprising: an obstacle data determination module, configured to quantitatively analyze the collision avoidance requirement information using an obstacle envelope method to obtain obstacle envelope data; an infeasible state domain calculation module, configured to calculate an infeasible state domain of a conventional operation of the obstacle envelope data based on a vehicle dynamics model; A safety analysis and judgment module is used to perform a safety analysis based on the conventional operation infeasible state domain, so as to determine whether to initiate an extreme collision avoidance operation according to the safety analysis result; The autonomous decision-making control module is used to determine whether to initiate extreme collision avoidance operations. It performs autonomous decision-making control of the vehicle based on the extreme dynamics control input data generated by the chassis domain control platform to achieve drift collision avoidance.

7. The system according to claim 6, wherein the obstacle data determination module is further configured to: Identify the location and shape of obstacles in the vehicle's surrounding environment based on the fusion results of the vehicle's lidar raw data and the Internet of Vehicles data; Determining a minimum circular envelope that completely encloses the detected obstacles based on the obstacle position and obstacle shape; Identify the minimum circular envelope to obtain obstacle envelope data; wherein, The obstacle envelope data includes the center coordinates and radius of the envelope.

8. The system according to claim 6 or 7, wherein the infeasible state domain calculation module is further configured to: establishing a vehicle dynamics model, and determining vehicle motion data based on the vehicle dynamics model; generating restriction conditions for vehicle status and action of conventional operations based on the vehicle motion data and preset road surface data; An optimal control problem is constructed based on the constraint conditions, the obstacle envelope data, and the current vehicle speed, and a numerical method is used to solve the optimal control problem to generate an infeasible state domain for conventional operation.

9. The system according to any one of claims 6 to 8, wherein the autonomous decision-making control module is further configured to: The extreme dynamics control input data generated by the chassis domain control platform is added to the conventional control term as a residual term, and the degree of intervention of the extreme dynamics control is judged based on the strength of the residual term. Using vehicle data under high side deviation, calibrate multiple sets of model parameters under high side deviation and normal conditions, and set switching rules; and Use rule-based, model-based or data-driven autonomous decision-making control algorithms to generate The corresponding optimization objectives are the collision avoidance performance index and the degree of intervention index on the original driving goal.

10. An electronic and electrical architecture for intelligent chassis drift and collision avoidance control using the method of claim 1, comprising: Chassis domain control platform, in-vehicle computing platform, Internet of Vehicles control platform, vehicle execution system, vehicle sensing system, and Gigabit Ethernet in-vehicle transmission system; The chassis domain control platform is used to coordinate and control the vehicle's power system, braking system, and suspension system, as well as autonomous decision-making and control under extreme working conditions; The vehicle-mounted computing platform is used to process raw data from various sensors and cameras; The Internet of Vehicles control platform is used for external communication; The vehicle sensing system is used to transmit raw data to the chassis domain control platform via the Gigabit Ethernet vehicle transmission system; The chassis domain control platform analyzes and responds to the raw data, shields other active safety functions and stability functions of the vehicle, and sends collision avoidance action instructions to the vehicle execution system; The vehicle execution system is used to perform drift collision avoidance in response to the collision avoidance action instruction.

Images