A steady state drift control method and system

Abstract

This invention discloses a steady-state drift control method and system, belonging to the field of vehicle dynamics technology. The method includes: constructing a set of vehicle dynamics equations comprising a three-degree-of-freedom vehicle dynamics model and a magic formula tire model; solving for the steady-state drift equilibrium point and outputting the longitudinal vehicle speed, lateral vehicle speed, yaw rate, and the front wheel steering angle and rear wheel slip ratio at the equilibrium point as control input references; training the system based on the control input references and the vehicle's real-time motion state by designing phased instantaneous reward functions and terminal reward functions that include a drift initiation phase and a steady-state drift maintenance phase, outputting a front wheel steering angle control command and a target value for the rear wheel slip ratio; and performing closed-loop tracking control of the target slip ratio through a slip ratio controller, outputting a driving and braking torque execution command to achieve steady-state drift control. This invention fully considers the nonlinear characteristics of the tires and the longitudinal and lateral coupling characteristics of vehicle dynamics, improving the accuracy and stability of steady-state drift control.

Description

Technical Field

[0001] This invention relates to the field of vehicle dynamics, and more particularly to a steady-state drift control method and system. Background Technology

[0002] In vehicle dynamics, drifting refers to a driving technique where a driver uses precise steering, throttle, and braking control to actively maintain the vehicle's sideslip motion while the rear wheels are at their tire grip limits. Essentially, it involves maintaining a high tire slip ratio to achieve controllable lateral movement within the constraints of the friction ellipse. Based on whether the equilibrium point at the grip limit changes, drifting can be classified into steady-state drifting and transient drifting.

[0003] Under steady-state drift conditions, the tire force reaches saturation and exhibits nonlinear characteristics. The longitudinal and lateral forces are highly coupled, making it difficult to solve for the equilibrium point and to accurately control the vehicle dynamics.

[0004] Patent CN 113928311 B proposes a closed-loop switching control method for steady-state vehicle drift. First, the steady-state drift equilibrium point is solved based on a three-degree-of-freedom vehicle dynamics model. Then, the steady-state drift process is divided into two parts: transition drift and steady-state drift maintenance. The transition drift controller uses longitudinal and lateral decomposition feedback control, while the steady-state drift controller uses an LQR controller. Patent CN 118025171 A proposes a drift control system combining automatic and assisted drift. First, a safety assessment is performed on the vehicle's dynamic state. When stability is good, the fully automatic control unit takes over drift control; when the vehicle is unstable, the auxiliary control unit assists the driver in drifting. Patent CN 113401112 A proposes a method to restore vehicle controllability through steady-state drift under unstable conditions such as large sideslip angles.

[0005] Summarizing existing vehicle steady-state drift control methods, it can be found that their drift equilibrium point solution relies on a simplified tire model, which does not adequately consider the longitudinal and lateral coupling and nonlinear characteristics of the tire, and directly uses the longitudinal force of the ground as the control input, which is difficult to implement in actual control. Their steady-state drift control relies on feedback control after the equilibrium point is linearized, but both the equilibrium point solution and linearization processes are affected by model errors, making it difficult to enter and maintain the steady-state drift state. Summary of the Invention

[0006] This invention proposes a steady-state drift control method that fully considers the nonlinear characteristics of the tire and the longitudinal and lateral coupling characteristics of vehicle dynamics, thereby improving the accuracy, stability and robustness of steady-state drift control.

[0007] Another objective of this invention is to provide a steady-state drift control system.

[0008] A third objective of this invention is to provide a non-transitory computer-readable storage medium.

[0009] To achieve the above objectives, a first aspect of the present invention provides a steady-state drift control method, comprising:

[0010] Based on vehicle structural parameters and tire characteristic parameters, a set of vehicle dynamic equations is constructed, including a three-degree-of-freedom vehicle dynamics model and a magic formula tire model, and the mapping relationship between tire longitudinal force, lateral force and tire slip angle and slip ratio is output. Based on the mapping relationship and the longitudinal vehicle speed and front wheel steering angle constraints at the target equilibrium point, the steady-state drift equilibrium point is solved, and the longitudinal vehicle speed, lateral vehicle speed, yaw rate, and the front wheel steering angle and rear wheel slip ratio at the equilibrium point are output as control input references. Based on the control input reference and the real-time motion state of the vehicle, and by designing a phased instantaneous reward function and a terminal reward function that includes the drift initiation stage and the steady-state drift maintenance stage, the system outputs the front wheel steering angle control command and the rear wheel slip ratio target value. Based on the front wheel steering angle control command, the rear wheel slip ratio target value, and the feedback of the vehicle's actual motion state, the slip ratio controller performs closed-loop tracking control of the target slip ratio and outputs the driving and braking torque execution command to achieve steady-state drift control of the vehicle.

[0011] In one embodiment of the present invention, a set of vehicle dynamic equations, including a three-degree-of-freedom vehicle dynamics model and a magic formula tire model, is constructed based on vehicle structural parameters and tire characteristic parameters. The mapping relationship between tire longitudinal force, lateral force, tire slip angle, and slip ratio is output, including: Based on the vehicle body mass, moment of inertia about the center of mass, distance from the front and rear axles to the center of mass, and front wheel steering angle, a three-degree-of-freedom vehicle dynamics model is established, which includes longitudinal motion equations, lateral motion equations, and yaw motion equations. Using a three-degree-of-freedom vehicle dynamics model, the longitudinal vehicle speed, lateral vehicle speed, yaw rate, and front wheel steering angle are input, and the tire slip angles of the front and rear axles are calculated based on the distances from the front and rear axles to the center of gravity. For rear-wheel drive, front-wheel steering vehicles, the front wheel slip ratio is set to zero and longitudinal slip occurs only in the rear wheels. A magic formula tire model with combined slip is adopted. The front and rear axle tire slip angles and slip ratios are used as inputs to calculate the longitudinal and lateral forces of the front and rear axle tires and output the mapping relationship between the tire longitudinal force, lateral force and tire slip angle and slip ratio.

[0012] In one embodiment of the present invention, based on the mapping relationship and the longitudinal vehicle speed and front wheel steering angle constraints at the target equilibrium point, the steady-state drift equilibrium point is solved, and the longitudinal vehicle speed, lateral vehicle speed, yaw rate, and the front wheel steering angle and rear wheel slip ratio at the equilibrium point, which serve as control input references, are output, including: Based on the longitudinal vehicle speed and front wheel steering angle constraints at the target equilibrium point, and setting the longitudinal vehicle speed, lateral vehicle speed and yaw rate to remain constant under steady-state drift conditions; Based on the mapping relationship, the longitudinal and lateral tire forces are expressed as corresponding functions of the tire slip angle and slip ratio; Based on longitudinal vehicle speed, lateral vehicle speed, yaw rate and vehicle structural parameters, calculate the tire slip angles of the front and rear axles; Substitute the tire slip angle and the set front wheel slip ratio of zero and the rear wheel slip ratio to be solved into the corresponding function, and solve the differential equations of the vehicle's longitudinal, lateral and yaw motions simultaneously to numerically solve for the longitudinal speed, lateral speed, yaw rate and rear wheel slip ratio. The outputs the longitudinal vehicle speed, lateral vehicle speed, and yaw rate at the equilibrium point obtained from the solution, as well as the front wheel steering angle constraint value and the rear wheel slip ratio solution value, which serve as the control input reference.

[0013] In one embodiment of the present invention, based on the control input reference and the real-time motion state of the vehicle, and through training by designing a phased instantaneous reward function and a terminal reward function that includes a drift initiation phase and a steady-state drift maintenance phase, the front wheel steering angle control command and the rear wheel slip ratio target value are output, including: Based on the longitudinal vehicle speed, lateral vehicle speed, yaw rate, front wheel steering angle and rear wheel slip ratio at the equilibrium point as reference states, a state space is constructed that includes the vehicle's longitudinal speed deviation, lateral speed deviation, yaw rate deviation and vehicle stage state flags. The structure includes a motion space that incorporates front wheel steering angle adjustment and rear wheel slip ratio adjustment. Design a phased, real-time reward function; Design a terminal reward function to give corresponding rewards or penalties based on whether the vehicle successfully enters and maintains a steady-state drift state at the end of the round. Based on the state space, action space, phased real-time reward function and terminal reward function, the TD3 algorithm, which includes a dual evaluation network and a corresponding target network, is used for training. The policy is updated by the minimum Q value and noise is added for exploration. The policy network is updated with a delay and the optimal policy network after training is output. The vehicle's real-time motion state is input into the optimal strategy network, and the front wheel steering angle control command and the rear wheel slip ratio target value are output.

[0014] In one embodiment of the present invention, the phased instant reward function includes an instant reward for the drift initiation phase designed based on the vehicle's rear wheel slip ratio, yaw rate, and sideslip angle, used to induce the vehicle to quickly enter a drift state with high slip, high sideslip, and high yaw rate; and an instant reward for the steady-state drift maintenance phase designed based on the deviation between the current state and the reference state, used to maintain the vehicle's state within a small range of fluctuations near the equilibrium point.

[0015] In one embodiment of the present invention, based on the front wheel steering angle control command, the rear wheel slip ratio target value, and feedback on the actual vehicle motion state, a slip ratio controller performs closed-loop tracking control of the target slip ratio, outputs a driving and braking torque execution command, and realizes steady-state drift control of the vehicle, including: The front wheel steering angle control command is directly used as the vehicle steering execution command; The target value of rear wheel slip ratio is used as the control target, and the actual longitudinal speed of the vehicle and the wheel speed are input to calculate the actual rear wheel slip ratio; Based on the deviation between the target value of the rear wheel slip ratio and the actual rear wheel slip ratio, the rear axle driving braking torque is calculated using an integral sliding mode control algorithm; Based on the steering execution command and the rear axle driving braking torque to the vehicle actuator, the rear wheels of the vehicle are driven into or maintained in a high slip state, thereby achieving control of the initiation phase or the maintenance phase of steady-state drift.

[0016] In one embodiment of the present invention, a terminal reward function is designed to provide corresponding rewards or penalties based on whether the vehicle successfully enters and maintains a steady-state drift state at the end of the round, including: Set vehicle stage status flags, including drift initiation stage, steady drift maintenance stage, rollover state and other abnormal states; If the vehicle is in a steady-state drift maintenance phase at the end of the round, it is considered to have successfully entered and maintained a steady-state drift state, and a high positive reward is given. If the vehicle is in the drifting phase at the end of the round, it is judged as failing to successfully enter a steady drift state and is given a moderate negative penalty; If the vehicle is overturned or in another abnormal state at the end of the round, the mission is considered a failure and a high negative penalty is imposed.

[0017] In one embodiment of the present invention, based on steering execution commands and rear axle driving braking torque to the vehicle actuators, the rear wheels of the vehicle are driven into or maintain a high slip state, thereby achieving drift initiation phase control or steady-state drift maintenance phase control, including: Input the vehicle stage status flag to determine the current stage of the vehicle; When it is determined to be the drift stage, a large front wheel angle is applied based on the steering execution command, and a large longitudinal force is applied based on the rear axle driving braking torque to drive the rear wheels into a high slip state and induce the vehicle into a side slip state. When the steady-state drift maintenance phase is determined, the front wheel angle is reduced based on the steering execution command to suppress excessive vehicle yaw, and the rear wheel high slip state is maintained based on the rear axle driving braking torque. The vehicle state is controlled to fluctuate in a small range near the steady-state drift equilibrium point to maintain the normal outward lateral speed. The front wheel steering angle and the rear axle driving braking torque are output to the corresponding actuators to achieve steady-state drift control at the corresponding stage.

[0018] To achieve the above objectives, a second aspect of the present invention provides a steady-state drift control system, comprising: The dynamic model construction module is configured to construct a set of vehicle dynamic equations, including a three-degree-of-freedom vehicle dynamic model and a magic formula tire model, based on vehicle structural parameters and tire characteristic parameters, and output the mapping relationship between tire longitudinal force, lateral force and tire slip angle and slip ratio. The steady-state drift equilibrium point solving module is configured to solve the steady-state drift equilibrium point based on the mapping relationship and the longitudinal vehicle speed and front wheel steering angle constraints of the target equilibrium point, and output the longitudinal vehicle speed, lateral vehicle speed, yaw rate at the equilibrium point, as well as the front wheel steering angle and rear wheel slip ratio as control input references. The TD3 reinforcement learning control module is configured to be trained based on the control input reference and the real-time motion state of the vehicle, and by designing a phased instantaneous reward function and a terminal reward function that includes a drift initiation phase and a steady-state drift maintenance phase, and outputs the front wheel steering angle control command and the rear wheel slip ratio target value. The slip ratio tracking control module is configured to perform closed-loop tracking control of the target slip ratio through the slip ratio controller based on the front wheel steering angle control command, the target value of the rear wheel slip ratio, and the feedback of the actual motion state of the vehicle, and output the drive and braking torque execution command to achieve steady-state drift control of the vehicle.

[0019] The steady-state drift control method and system of this invention first constructs a vehicle dynamics model and a magic formula tire model to solve for the steady-state drift equilibrium point. This method transforms the inputs to steady-state drift control into front wheel steering angle and rear wheel slip ratio. Then, the steady-state drift process is divided into drift initiation and drift maintenance phases. A reinforcement learning controller based on the TD3 reinforcement learning method is designed accordingly. By designing phased immediate and final reward functions, the agent achieves optimal steady-state drift control output through interaction with the environment. Finally, a slip ratio controller is used to achieve precise control of vehicle dynamics, realizing the initiation of steady-state drift and maintenance of the critical stable state. This invention introduces slip ratio control into the solution of the steady-state drift equilibrium point and steady-state drift control, improving the accuracy of steady-state drift control. Furthermore, it does not rely on additional onboard sensors and actuators, making it highly valuable for engineering applications.

[0020] To achieve the above objectives, a third aspect of this application provides a non-transitory computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steady-state drift control method as described in the first aspect embodiment.

[0021] The beneficial effects of this invention are as follows: This invention introduces a magic formula tire model into the solution of steady-state drift equilibrium point. Compared with existing methods for solving steady-state drift equilibrium point, the tire model is more accurate and fully considers the tire's nonlinear characteristics and longitudinal and lateral coupling characteristics under the adhesion limit condition of drift, thus improving the rationality and accuracy of the steady-state drift equilibrium point solution method.

[0022] Inspired by the phased control method for entering and maintaining steady-state drift, a steady-state drift control method based on TD3 reinforcement learning is proposed. The current vehicle motion state is considered in the design of the reward function, which avoids the problems of insufficient accuracy and poor stability caused by model bias and environmental disturbances in traditional feedback control methods after the equilibrium point is linearized.

[0023] This invention introduces slip ratio and slip ratio control into the solution of steady-state drift equilibrium point and steady-state drift control, using the tire longitudinal slip ratio instead of the ground force as the control input, thereby improving the accuracy and feasibility of steady-state drift control.

[0024] The methods of this invention are all based on commonly used chassis sensors, which do not require the introduction of additional vehicle sensors, resulting in low hardware development costs and strong practicality.

[0025] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0026] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A flowchart of a steady-state drift control method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the main modules of the steady-state drift control method architecture provided in the embodiments of the present invention; Figure 3 This is a structural diagram of a steady-state drift control system provided in an embodiment of the present invention. Detailed Implementation

[0027] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0028] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0029] A steady-state drift control method and system according to an embodiment of the present invention will now be described with reference to the accompanying drawings.

[0030] This embodiment provides a steady-state drift control method. For example... Figure 1 As shown, it includes: S1, based on vehicle structural parameters and tire characteristic parameters, constructs a set of vehicle dynamic equations including a three-degree-of-freedom vehicle dynamics model and a magic formula tire model, and outputs the mapping relationship between tire longitudinal force, lateral force and tire slip angle and slip ratio. S2, based on the mapping relationship and the longitudinal vehicle speed and front wheel steering angle constraints at the target equilibrium point, solve for the steady-state drift equilibrium point, and output the longitudinal vehicle speed, lateral vehicle speed, yaw rate at the equilibrium point, as well as the front wheel steering angle and rear wheel slip ratio as control input references; S3, based on the control input reference and the real-time motion state of the vehicle, is trained by designing a phased instantaneous reward function and a terminal reward function that includes the drift initiation stage and the steady-state drift maintenance stage, and outputs the front wheel steering angle control command and the rear wheel slip ratio target value. S4, based on the front wheel steering angle control command, the rear wheel slip ratio target value, and the feedback of the vehicle's actual motion state, performs closed-loop tracking control of the target slip ratio through the slip ratio controller, outputs the driving and braking torque execution command, and realizes the vehicle's steady-state drift control.

[0031] The steady-state drift control method of this invention incorporates the magic formula tire model into the dynamic equations for solving the steady-state drift equilibrium point. Based on a three-degree-of-freedom vehicle dynamics model and a high-precision tire model, it accurately solves for the steady-state drift equilibrium point. The method's architecture is as follows: Figure 2 As shown, the method may specifically include the following steps: Establish a three-degree-of-freedom vehicle dynamics model in the vehicle body coordinate system: (1) In the formula, Indicates longitudinal vehicle speed. Indicates lateral vehicle speed. Indicates yaw rate. For the front wheel steering angle, , These represent the distances from the front and rear axles to the center of mass, respectively. and This represents the vehicle's mass and moment of inertia about its center of mass.

[0032] For steady-state drift conditions, considering a rear-wheel-drive, front-wheel-steering vehicle, longitudinal slip is assumed to occur only in the rear wheels, with a front wheel slip ratio of 0. A magic formula tire model based on combined slip is adopted, using the tire slip angle... and slip ratio As input, calculate the longitudinal force of the tire. With lateral force .

[0033] (2) Furthermore, according to the magic formula tire model, the longitudinal and lateral tire forces can be considered as a bivariate function of slip ratio and tire slip angle, i.e.

[0034] The formulas for the tire slip angles of the front and rear axles are as follows: (3) In steady-state drifting, the vehicle drifts along a trajectory with a certain curvature while maintaining a constant attitude. The vehicle's longitudinal speed, lateral speed, and yaw rate remain constant in the vehicle's coordinate system. That is: (4) Substituting formula (4) into the vehicle dynamics equations, after simplification, we have a total of Five unknowns, set the equilibrium point speed and front wheel steering angle inputs. This allows us to treat the states of the other three equilibrium points as unknowns, solve the differential equation numerically, and obtain the longitudinal vehicle speed, lateral vehicle speed, yaw rate, front wheel steering angle input, and rear wheel slip input at the equilibrium point.

[0035] Based on the above methods for solving the steady-state drift equilibrium point, and referring to expert experience, a reinforcement learning-based steady-state drift control method is designed. Considering that in actual steady-state drifting operations by racing drivers, there are two phases: drift entry and drift maintenance. In the drift entry phase, i.e., the drift initiation phase, the driver applies a large longitudinal force to induce a high-slip rear axle, while simultaneously applying a large front wheel steering angle to induce a sideslip. In the drift maintenance phase, the front wheel steering angle is reduced to suppress excessive yaw and maintain a normal outward lateral velocity, keeping the vehicle's dynamic state and control input near the steady-state drift equilibrium point, thereby maintaining this dynamic critical stable state of steady drift.

[0036] Therefore, a steady-state drift control method based on TD3 reinforcement learning is designed. First, the state space and action space are defined.

[0037] (5) in, The markers indicate the vehicle's status, including four states: the initial drift stage, the steady drift maintenance stage, rollover, and other states.

[0038] To reflect the different stages of the steady-state drift process, this invention divides the drift process into the initial drift stage and the steady-state drift maintenance stage, and designs different instantaneous reward functions accordingly. At the same time, it designs a terminal reward to reflect whether the task has been successfully completed.

[0039] The reward function design considers both the final reward and the immediate reward. Immediate rewards are designed for both the initial drift phase and the steady-state drift phase: In the initial drift phase, the vehicle is expected to quickly enter a drift state with high slip, high sideslip, and high yaw rate; therefore, the immediate reward can be designed as follows: (6) During the steady-state drift maintenance phase, it is desirable for the vehicle state and control input to fluctuate within a small range around the equilibrium point. Therefore, the immediate reward can be designed to account for the deviation from the equilibrium point. (7) At the end of the round, a terminal reward is given based on the final state: (8) A high reward is given for successfully entering and maintaining a steady-state drift, while other states are penalized to varying degrees. The final reward function is: (9) In summary, the total reward obtained by the agent during the entire training process is: (10) In the formula, This is the discount factor.

[0040] TD3 reinforcement learning is a classic action-evaluation reinforcement learning method. The algorithm includes an action neural network. Two evaluation neural networks , and the corresponding target neural network , and .

[0041] Its core features are: Two evaluation networks and two target evaluation networks are employed, and a minimum Q-value update strategy is used to suppress the overestimation of the value function and improve the reliability of the policy gradient estimation. (11) Adding pruning noise to the actions output by the policy prevents the exploration process from being limited to extreme action regions, thereby improving the policy's generalization ability in the continuous action space. The action update can be represented as: (12) By adopting a delayed update policy network approach, the action neural network is updated only after multiple evaluation networks have been updated, thus avoiding the instability caused by frequent updates to gradients and action outputs, thereby improving the stability and convergence of action outputs during policy training.

[0042] TD3 has significant advantages in stability and convergence speed in continuous control problems. It is suitable for nonlinear dynamic control scenarios, including steady-state drift, and can effectively maintain the robustness and safety of the strategy in high-dimensional continuous actions, thereby ensuring that the control strategy still has high reliability and engineering feasibility under extreme attachment conditions.

[0043] In the reinforcement learning training and subsequent result verification phases, considering that the actual execution input of the vehicle is driving and braking torque, a slip ratio control module is designed to control the action input. Tracking control is performed. The slip ratio control module can take the form of an integral sliding mode controller, etc.

[0044] In summary, this invention first constructs a vehicle dynamics model and a magic formula tire model to solve for the steady-state drift equilibrium point. This method transforms the inputs to steady-state drift control into front wheel steering angle and rear wheel slip ratio, providing a foundation for precise steady-state drift control. Then, inspired by the experience of racing car drivers and rule-based controllers, the steady-state drift process is divided into drift initiation and drift maintenance phases. Based on this, a reinforcement learning controller based on the TD3 reinforcement learning method is designed. By designing phased immediate and final reward functions, the agent achieves optimal steady-state drift control outputs through interaction with the environment. Finally, a slip ratio controller is used to achieve precise control of vehicle dynamics, realizing steady-state drift initiation and critical stability maintenance, thus achieving steady-state drift control based on the reinforcement learning method.

[0045] To implement the method of the embodiments of the present invention, the present invention also proposes a steady-state drift control system 10, such as... Figure 3 As shown, it includes: The dynamic model construction module 100 is configured to construct a set of vehicle dynamic equations, including a three-degree-of-freedom vehicle dynamic model and a magic formula tire model, based on vehicle structural parameters and tire characteristic parameters, and output the mapping relationship between tire longitudinal force, lateral force and tire slip angle and slip ratio. The steady-state drift equilibrium point solving module 200 is configured to solve the steady-state drift equilibrium point based on the mapping relationship and the longitudinal vehicle speed and front wheel steering angle constraints of the target equilibrium point, and output the longitudinal vehicle speed, lateral vehicle speed, yaw rate at the equilibrium point, as well as the front wheel steering angle and rear wheel slip ratio as control input references. The TD3 reinforcement learning control module 300 is configured to be trained based on the control input reference and the real-time motion state of the vehicle, and by designing a phased instantaneous reward function and a terminal reward function that includes a drift initiation phase and a steady-state drift maintenance phase, and outputs the front wheel steering angle control command and the rear wheel slip ratio target value. The slip ratio tracking control module 400 is configured to perform closed-loop tracking control of the target slip ratio based on the front wheel steering angle control command, the target value of the rear wheel slip ratio, and the feedback of the actual motion state of the vehicle, and output the drive and braking torque execution command to achieve steady-state drift control of the vehicle.

[0046] The steady-state drift control system of this invention first constructs a vehicle dynamics model and a magic formula tire model to solve for the steady-state drift equilibrium point. This method transforms the inputs to steady-state drift control into front wheel steering angle and rear wheel slip ratio. Then, the steady-state drift process is divided into drift initiation and drift maintenance phases. A reinforcement learning controller based on the TD3 reinforcement learning method is designed accordingly. By designing phased immediate and final reward functions, the agent achieves optimal steady-state drift control output through interaction with the environment. Finally, a slip ratio controller is used to achieve precise control of vehicle dynamics, realizing the initiation of steady-state drift and maintenance of the critical stable state. This invention introduces slip ratio control into the solution of the steady-state drift equilibrium point and steady-state drift control, improving the accuracy of steady-state drift control without relying on additional onboard sensors and actuators, thus possessing significant engineering application value.

[0047] To implement the above embodiments, this application also proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the method described in the foregoing embodiments.

[0048] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0049] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0050] 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 indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A steady-state drift control method, characterized in that, include: Based on vehicle structural parameters and tire characteristic parameters, a set of vehicle dynamic equations is constructed, including a three-degree-of-freedom vehicle dynamics model and a magic formula tire model, and the mapping relationship between tire longitudinal force, lateral force and tire slip angle and slip ratio is output. Based on the mapping relationship and the longitudinal vehicle speed and front wheel steering angle constraints at the target equilibrium point, the steady-state drift equilibrium point is solved, and the longitudinal vehicle speed, lateral vehicle speed, yaw rate, and the front wheel steering angle and rear wheel slip ratio at the equilibrium point are output as control input references. Based on the control input reference and the real-time motion state of the vehicle, and by designing a phased instantaneous reward function and a terminal reward function that includes the drift initiation stage and the steady-state drift maintenance stage, the system outputs the front wheel steering angle control command and the rear wheel slip ratio target value. Based on the front wheel steering angle control command, the rear wheel slip ratio target value, and the feedback of the vehicle's actual motion state, the slip ratio controller performs closed-loop tracking control of the target slip ratio and outputs the driving and braking torque execution command to achieve steady-state drift control of the vehicle.

2. The method as described in claim 1, characterized in that, A set of vehicle dynamics equations is constructed based on vehicle structural parameters and tire characteristic parameters, including a three-degree-of-freedom vehicle dynamics model and a magic formula tire model. The mapping relationships between tire longitudinal force, lateral force, tire slip angle, and slip ratio are output, including: Based on the vehicle body mass, moment of inertia about the center of mass, distance from the front and rear axles to the center of mass, and front wheel steering angle, a three-degree-of-freedom vehicle dynamics model is established, which includes longitudinal motion equations, lateral motion equations, and yaw motion equations. Using a three-degree-of-freedom vehicle dynamics model, the longitudinal vehicle speed, lateral vehicle speed, yaw rate, and front wheel steering angle are input, and the tire slip angles of the front and rear axles are calculated based on the distances from the front and rear axles to the center of gravity. For rear-wheel drive, front-wheel steering vehicles, the front wheel slip ratio is set to zero and longitudinal slip occurs only in the rear wheels. A magic formula tire model with combined slip is adopted. The front and rear axle tire slip angles and slip ratios are used as inputs to calculate the longitudinal and lateral forces of the front and rear axle tires and output the mapping relationship between the tire longitudinal force, lateral force and tire slip angle and slip ratio.

3. The method as described in claim 1, characterized in that, Based on the mapping relationship and the longitudinal vehicle speed and front wheel steering angle constraints at the target equilibrium point, the steady-state drift equilibrium point is solved, and the longitudinal vehicle speed, lateral vehicle speed, yaw rate, and the front wheel steering angle and rear wheel slip ratio as control input references at the equilibrium point are output, including: Based on the longitudinal vehicle speed and front wheel steering angle constraints at the target equilibrium point, and setting the longitudinal vehicle speed, lateral vehicle speed and yaw rate to remain constant under steady-state drift conditions; Based on the mapping relationship, the longitudinal and lateral tire forces are expressed as corresponding functions of the tire slip angle and slip ratio; Based on longitudinal vehicle speed, lateral vehicle speed, yaw rate and vehicle structural parameters, calculate the tire slip angles of the front and rear axles; Substitute the tire slip angle and the set front wheel slip ratio of zero and the rear wheel slip ratio to be solved into the corresponding function, and solve the differential equations of the vehicle's longitudinal, lateral and yaw motions simultaneously to numerically solve for the longitudinal speed, lateral speed, yaw rate and rear wheel slip ratio. The outputs the longitudinal vehicle speed, lateral vehicle speed, and yaw rate at the equilibrium point obtained from the solution, as well as the front wheel steering angle constraint value and the rear wheel slip ratio solution value, which serve as the control input reference.

4. The method as described in claim 1, characterized in that, Based on the control input baseline and the vehicle's real-time motion state, and through training using a phased instantaneous reward function and a terminal reward function that includes a drift initiation phase and a steady-state drift maintenance phase, the system outputs front wheel steering angle control commands and rear wheel slip ratio target values, including: Based on the longitudinal vehicle speed, lateral vehicle speed, yaw rate, front wheel steering angle and rear wheel slip ratio at the equilibrium point as reference states, a state space is constructed that includes the vehicle's longitudinal speed deviation, lateral speed deviation, yaw rate deviation and vehicle stage state flags. The structure includes a motion space that incorporates front wheel steering angle adjustment and rear wheel slip ratio adjustment. Design a phased, real-time reward function; Design a terminal reward function to give corresponding rewards or penalties based on whether the vehicle successfully enters and maintains a steady-state drift state at the end of the round. Based on the state space, action space, phased real-time reward function and terminal reward function, the TD3 algorithm, which includes a dual evaluation network and a corresponding target network, is used for training. The policy is updated by the minimum Q value and noise is added for exploration. The policy network is updated with a delay and the optimal policy network after training is output. The vehicle's real-time motion state is input into the optimal strategy network, and the front wheel steering angle control command and the rear wheel slip ratio target value are output.

5. The method as described in claim 4, characterized in that, The phased instant reward function includes an instant reward for the drift initiation phase designed based on the vehicle's rear wheel slip ratio, yaw rate, and sideslip angle, used to induce the vehicle to quickly enter a drift state with high slip, high sideslip, and high yaw rate; and an instant reward for the steady-state drift maintenance phase designed based on the deviation between the current state and the reference state, used to maintain the vehicle's state within a small range of fluctuations around the equilibrium point.

6. The method as described in claim 1, characterized in that, Based on the front wheel steering angle control command, the rear wheel slip ratio target value, and feedback on the vehicle's actual motion state, the slip ratio controller performs closed-loop tracking control of the target slip ratio, outputting drive and braking torque execution commands to achieve steady-state drift control of the vehicle, including: The front wheel steering angle control command is directly used as the vehicle steering execution command; The target value of rear wheel slip ratio is used as the control target, and the actual longitudinal speed of the vehicle and the wheel speed are input to calculate the actual rear wheel slip ratio; Based on the deviation between the target value of the rear wheel slip ratio and the actual rear wheel slip ratio, the rear axle driving braking torque is calculated using an integral sliding mode control algorithm; Based on the steering execution command and the rear axle driving braking torque to the vehicle actuator, the rear wheels of the vehicle are driven into or maintained in a high slip state, thereby achieving control of the initiation phase or the maintenance phase of steady-state drift.

7. The method as described in claim 4, characterized in that, Design a terminal reward function that provides corresponding rewards or penalties based on whether the vehicle successfully enters and maintains a steady-state drift at the end of the round, including: Set vehicle stage status flags, including drift initiation stage, steady drift maintenance stage, rollover state and other abnormal states; If the vehicle is in a steady-state drift maintenance phase at the end of the round, it is considered to have successfully entered and maintained a steady-state drift state, and a high positive reward is given. If the vehicle is in the drifting phase at the end of the round, it is judged as failing to successfully enter a steady drift state and is given a moderate negative penalty; If the vehicle is overturned or in another abnormal state at the end of the round, the mission is considered a failure and a high negative penalty is imposed.

8. The method as described in claim 6, characterized in that, Based on steering commands and rear axle braking torque transmitted to the vehicle actuators, the rear wheels of the vehicle are driven into or maintained in a high-slip state, achieving drift initiation control or steady-state drift maintenance control, including: Input the vehicle stage status flag to determine the current stage of the vehicle; When it is determined to be the drift stage, a large front wheel angle is applied based on the steering execution command, and a large longitudinal force is applied based on the rear axle driving braking torque to drive the rear wheels into a high slip state and induce the vehicle into a side slip state. When the steady-state drift maintenance phase is determined, the front wheel angle is reduced based on the steering execution command to suppress excessive vehicle yaw, and the rear wheel high slip state is maintained based on the rear axle driving braking torque. The vehicle state is controlled to fluctuate in a small range near the steady-state drift equilibrium point to maintain the normal outward lateral speed. The front wheel steering angle and the rear axle driving braking torque are output to the corresponding actuators to achieve steady-state drift control at the corresponding stage.

9. A steady-state drift control system, characterized in that, include: The dynamic model construction module is configured to construct a set of vehicle dynamic equations, including a three-degree-of-freedom vehicle dynamic model and a magic formula tire model, based on vehicle structural parameters and tire characteristic parameters, and output the mapping relationship between tire longitudinal force, lateral force and tire slip angle and slip ratio. The steady-state drift equilibrium point solving module is configured to solve the steady-state drift equilibrium point based on the mapping relationship and the longitudinal vehicle speed and front wheel steering angle constraints of the target equilibrium point, and output the longitudinal vehicle speed, lateral vehicle speed, yaw rate at the equilibrium point, as well as the front wheel steering angle and rear wheel slip ratio as control input references. The TD3 reinforcement learning control module is configured to be trained based on the control input reference and the real-time motion state of the vehicle, and by designing a phased instantaneous reward function and a terminal reward function that includes a drift initiation phase and a steady-state drift maintenance phase, and outputs the front wheel steering angle control command and the rear wheel slip ratio target value. The slip ratio tracking control module is configured to perform closed-loop tracking control of the target slip ratio through the slip ratio controller based on the front wheel steering angle control command, the target value of the rear wheel slip ratio, and the feedback of the actual motion state of the vehicle, and output the drive and braking torque execution command to achieve steady-state drift control of the vehicle.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements a steady-state drift control method as described in any one of claims 1-8.

Images