Model prediction and reinforcement learning based rear wheel active steering control method and system

Abstract

The present disclosure provides a rear wheel active steering control method and system based on model prediction and reinforcement learning, relating to the technical field of automobile control, comprising: inputting the front wheel steering angle, vehicle speed, yaw rate and centroid side slip angle of the vehicle into a model predictive controller, taking the given reference trajectory as the control target, rolling optimization of the output of the model predictive controller, and outputting the optimal rear wheel steering control amount; inputting the vehicle state information and the optimal rear wheel steering control amount into a deep reinforcement learning network, taking a multi-layer fully connected neural network as the action network of the Actor in the deep deterministic policy gradient reinforcement learning algorithm, setting the observation state value and the reward, and using the deep deterministic policy gradient reinforcement learning algorithm to track the centroid side slip angle and the yaw rate target, so that the yaw rate and the centroid side slip angle are as close to the expected value as possible, and the optimal rear wheel steering control is achieved.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of automobile control, in particular to a rear wheel active steering control method and system based on model prediction and reinforcement learning. BACKGROUND

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute the prior art.

[0003] Vehicle handling stability is increasingly valued as an important performance of modern vehicles. Vehicle active safety technology can assist the driver's operation, so as to reduce and eliminate the probability of vehicle danger as much as possible, so effective improvement of the handling stability of the vehicle in dangerous conditions has been the focus of vehicle research for a long time. The main factors affecting vehicle handling stability are steering and stability control technology, including front wheel active steering, rear wheel active steering, active four-wheel steering, servo steering, and vehicle stability control technology. The above technical solutions are mainly realized through vehicle lateral force (steering control) and longitudinal force (drive and brake) control, and the typical ones include vehicle stability control technology and rear wheel active steering technology.

[0004] Rear wheel active steering is one of the methods to realize vehicle active safety, and its main advantage is the control of vehicle mass center side slip angle. When steering, the mass center side slip angle can be basically maintained at zero, which can greatly improve the dynamic response characteristics of the vehicle to the steering wheel input, and can improve the transient response of yaw rate and lateral acceleration. In addition, for long-wheelbase vehicles, the turning radius can be significantly reduced at low speed, improving the maneuverability of the vehicle at low speed. The existing rear wheel active steering method is a simple front-rear wheel proportional control or a feedback control according to the vehicle body yaw rate. The above methods cannot adaptively control according to the environment. SUMMARY

[0005] To solve the above problems, the present disclosure provides a rear wheel active steering control method and system based on model prediction and reinforcement learning. A simplified two-degree-of-freedom single-track model of the whole vehicle is used to design a model predictive controller. The dynamic state of the vehicle body and the output of the model predictive controller are obtained based on joint simulation. A multi-layer fully connected neural network is used as the action network of the Actor in the DDPG reinforcement learning algorithm. The DDPG algorithm is used for multi-condition learning of trajectory tracking and heading angle tracking in the CarSim platform. Finally, the DDPG agent algorithm trained based on multi-condition learning is used for real vehicle multi-condition learning, so as to achieve optimal comprehensive performance of active rear wheel steering control.

[0006] According to some embodiments, the present disclosure adopts the following technical solutions:

[0007] The rear wheel active steering control method based on model prediction and reinforcement learning comprises the following steps:

[0008] A two-degree-of-freedom vehicle single-track model is constructed, and a model predictive controller is designed based on the two-degree-of-freedom vehicle single-track model.

[0009] The front wheel steering angle, vehicle speed, yaw rate and center of mass side slip angle of the vehicle are input into the model predictive controller, a given reference trajectory is taken as a control target, the output of the model predictive controller is optimized, and an optimal rear wheel steering control amount is obtained.

[0010] The vehicle state information and the optimal rear wheel steering control amount are input into a deep reinforcement learning network, a multi-layer fully connected neural network is taken as an action network of an Actor in a deep deterministic policy gradient reinforcement learning algorithm, an expected value is given, an observation state value and a reward are set, the deep deterministic policy gradient reinforcement learning algorithm is used for target tracking of the center of mass side slip angle and the yaw rate, the yaw rate and the center of mass side slip angle are made to be as close as possible to the expected value, and optimal rear wheel steering control is realized.

[0011] According to some embodiments, the present disclosure adopts the following technical solution:

[0012] The rear wheel active steering control system based on model prediction and reinforcement learning comprises:

[0013] A model construction module is configured to construct a two-degree-of-freedom vehicle single-track model, and design a model predictive controller based on the two-degree-of-freedom vehicle single-track model.

[0014] A prediction control module is configured to input the front wheel steering angle, vehicle speed, yaw rate and center of mass side slip angle of the vehicle into the model predictive controller, take a given reference trajectory as a control target, optimize the output of the model predictive controller, and obtain an optimal rear wheel steering control amount.

[0015] An active steering control module is configured to input the vehicle state information and the optimal rear wheel steering control amount into a deep reinforcement learning network, take a multi-layer fully connected neural network as an action network of an Actor in a deep deterministic policy gradient reinforcement learning algorithm, give an expected value, set an observation state value and a reward, use the deep deterministic policy gradient reinforcement learning algorithm for target tracking of the center of mass side slip angle and the yaw rate, make the yaw rate and the center of mass side slip angle as close as possible to the expected value, and realize optimal rear wheel steering control.

[0016] According to some embodiments, the present disclosure adopts the following technical solution:

[0017] A computer program product comprises a computer program, which, when executed by a processor, implements the rear wheel active steering control method based on model prediction and reinforcement learning.

[0018] According to some embodiments, the present disclosure adopts the technical solutions as follows:

[0019] A non-transitory computer readable storage medium for storing computer instructions, the computer instructions being executed by a processor to implement the rear wheel active steering control method based on model prediction and reinforcement learning.

[0020] According to some embodiments, the present disclosure adopts the technical solutions as follows:

[0021] An electronic device comprising a processor, a memory and a computer program; wherein the processor is connected with the memory, and the computer program is stored in the memory; when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the rear wheel active steering control method based on model prediction and reinforcement learning.

[0022] Compared with the prior art, the present disclosure has the beneficial effects that:

[0023] The rear wheel active steering control method based on model prediction and reinforcement learning of the present disclosure takes the whole vehicle system as the research object, firstly designs a model prediction controller by taking a simplified two-degree-of-freedom single-track model of the whole vehicle as a prediction model, obtains the dynamic state of the vehicle body and the output of the model prediction controller based on Simulink / C arSim joint simulation, and uses the same as a data set to train a multi-layer fully connected neural network; then uses the trained neural network as the action network of the Actor in the reinforcement learning algorithm DDPG, and uses the DDPG algorithm to learn track tracking, heading angle tracking and other tasks in the CarSim platform; finally, the DDPG intelligent agent algorithm trained based on multi-condition learning is used for real vehicle multi-condition learning, so as to achieve an active rear wheel steering control system with optimal comprehensive performance. BRIEF DESCRIPTION OF DRAWINGS

[0024] The accompanying drawings, which form a part of this disclosure, are intended to provide further understanding of the present disclosure and are incorporated herein for illustrative purposes. The schematic embodiments of the present disclosure and their descriptions are used to explain the present disclosure, and do not constitute an improper limitation on the present disclosure.

[0025] Figure 1 A model prediction control structure schematic diagram of the embodiment of the present disclosure;

[0026] Figure 2 An Actor Network structure diagram of the embodiment of the present disclosure;

[0027] Figure 3 A Critic Network structure diagram of the embodiment of the present disclosure

[0028] Figure 4Flowchart of the reinforcement learning algorithm of an embodiment of the present disclosure. DETAILED DESCRIPTION

[0029] The present disclosure is further described below with reference to the accompanying drawings and examples.

[0030] It should be noted that the following detailed description is exemplary in nature and is intended to provide further description of the present disclosure. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this present disclosure belongs.

[0031] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments in accordance with the present disclosure. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, steps, operations, elements, components, and / or groups thereof, but do not preclude the presence or addition of one or more other features, steps, operations, elements, components, and / or groups thereof.

[0032] Example 1

[0033] In an embodiment of the present disclosure, a rear wheel active steering control method based on model prediction and reinforcement learning is provided, comprising:

[0034] Step 1: a two-degree-of-freedom vehicle single-track model is constructed, and a model predictive controller is designed based on the two-degree-of-freedom vehicle single-track model;

[0035] Step 2: the front wheel steering angle, vehicle speed, yaw rate and center of mass side slip angle of the vehicle are input into the model predictive controller, a given reference trajectory is taken as a control target, and the output of the rolling optimization model predictive controller is obtained to obtain the optimal rear wheel steering angle control amount;

[0036] Step 3: the vehicle state information and the optimal rear wheel steering angle control amount are input into a deep reinforcement learning network, a multi-layer fully connected neural network is taken as an action network of an Actor in a deep deterministic policy gradient reinforcement learning algorithm, an expected value is given, an observation state value and a reward are set, and the deep deterministic policy gradient reinforcement learning algorithm is used for target tracking of the center of mass side slip angle and the yaw rate, so that the yaw rate and the center of mass side slip angle are as close to the expected value as possible, and optimal rear wheel steering control is achieved.

[0037] As an embodiment, the specific implementation process of the rear wheel active steering control method based on model prediction and reinforcement learning of the present disclosure is as follows:

[0038] Step 1: a two-degree-of-freedom vehicle single-track model is constructed, and a model predictive controller is designed based on the two-degree-of-freedom vehicle single-track model;

[0039] Specifically, a two-degree-of-freedom vehicle single-track model is constructed, including:

[0040]

[0041] Take X(t) = [β γ] T , u(t) = δ r , ω(t) = δ f , Y = γ

[0042] The dynamic model matrix can be obtained as follows:

[0043]

[0044] In the formula,

[0045]

[0046] Wherein, k f , k r are the front and rear wheel cornering stiffness respectively; L f , L r are the distances from the front axle / rear axle to the center of mass, m is the mass of the vehicle, v is the longitudinal speed of the vehicle, I z is the rotational inertia of the vehicle around the z-axis, β is the center of mass cornering angle, gamma is the yaw rate, δ f is the front wheel steering angle, and δ r is the rear wheel steering angle.

[0047] Further, a model predictive controller is designed based on the two-degree-of-freedom vehicle single-track model, including: discretizing the continuous equation to obtain a new discrete state space model, which is,

[0048]

[0049] In the formula, the vector x is the state vector, the matrix A is the system matrix, the matrix B is the control matrix, the matrix F is the disturbance matrix, the vector u is the control vector, the vector ω is the disturbance vector, and the vector Y is the output vector. The matrix C is the output matrix.

[0050] A = (I + T s ) A c , B = T s B u , F = T s B ω

[0051] Wherein, T s is the sampling period of system discretization, and I is the unit matrix.

[0052] Let,

[0053]

[0054] A new discrete state space model is obtained,

[0055]

[0056] wherein,

[0057]

[0058] Further, in the predicted p-step time domain, the prediction equation is constructed as,

[0059] Y p (k+1|k) = S x Z(k) + S u ΔU(k)

[0060] wherein,

[0061]

[0062] Further, given the reference quantity, the objective function is constructed, the reference quantity is Taking the front wheel steering angle, vehicle speed, yaw rate and centroid side slip angle of the vehicle as input, and taking the given reference trajectory as control target, the output prediction value of the rolling optimization and constantly correcting model predictive controller is optimized, and the objective function is

[0063] J = ||Γ y (Y p (k+1|k) - R(k+1))|| 2

[0064] +‖Γ u ΔU(k)‖ 2

[0065] +ρε 2 +‖Γ u ΔU(k)‖ 2 +ρε 2 #(20)

[0066] wherein, Γ y , Γ u are the weighting matrices of the reference quantity and the control quantity respectively, ρ is the weight coefficient, ε is the relaxation factor, and R(k+1) is the given control output reference sequence, which is

[0067]

[0068] The objective function and the corresponding control constraint condition constitute a quadratic programming problem, which is solved by a solver to obtain the optimal rear wheel steering angle control quantity δ r .

[0069] Step 2: input the vehicle state information and the optimal rear wheel steering control amount into the deep reinforcement learning network, use a multi-layer fully connected neural network as the action network of the Actor in the deep deterministic policy gradient reinforcement learning algorithm, set the observation state value and the reward given the expected value, and use the deep deterministic policy gradient reinforcement learning algorithm to track the target of the centroid side slip angle and the yaw rate, so that the yaw rate and the centroid side slip angle are as close to the expected value as possible, and the optimal rear wheel steering control is achieved.

[0070] Specifically, step 1) collects vehicle state information based on Simulink / CarSim joint simulation data acquisition, and collects the output of the MPC controller and the vehicle state information in the CarSim vehicle state information under different vehicle speeds, road adhesion coefficients, and road profiles.

[0071] 2) build a multi-layer fully connected neural network in Matlab, and use the output of the MPC controller and the vehicle state information in the CarSim vehicle state information for pre-training; the pre-training process: collect the control sequence of the MPC controller, use these sequences as a data set, and pre-train the Actor network in the deep reinforcement learning through deep learning (imitation learning).

[0072] Further, the DDPG (deep deterministic policy gradient reinforcement learning algorithm) agent and the learning environment are constructed, including:

[0073] DDPG algorithm is a deep reinforcement learning algorithm based on policy gradient, which can be used to solve problems in continuous action space. In continuous action space, the agent can take any numerical action. It is based on Actor-Critic, uses policy network μ(s; θ) for decision output, and uses value network Q(s, a; ω) to score the decisions executed by the policy network, and then guide the policy network to improve, and also based on the extension of DQN algorithm, uses experience replay and target network to stabilize the training process.

[0074] In the training of the policy network, DDPG uses deterministic policy gradient DPG to make unbiased estimate

[0075]

[0076] Thus, the algorithm for updating θ is obtained. Each time a state is randomly extracted from the experience replay buffer, denoted as s j , calculate a j = μ(s j ; θ), and update θ using the gradient ascent formula:

[0077]

[0078] Here, β is the learning rate, which generally needs to be adjusted manually. Gradient ascent can make the objective function J(θ) increase, that is, make the value network score higher for the action made by the decision network. In the training of the value network, similar to DQN, the TD algorithm is used to make the prediction of the value network closer to the TD target, and the prediction value of the target network closer to the true value function, that is,

[0079]

[0080] The gradient thereof is:

[0081]

[0082] The update formula of the value network is:

[0083]

[0084] In reinforcement learning, the algorithm using a single "Q neural network" can cause the learning process to be unstable. Because the parameters of the value network are constantly modified during training, and are also used in the calculation of the decision network. This mixed use can cause large variance and instability in the training process, making it difficult for the algorithm to converge to the optimal solution. In order to provide stable estimates during training and improve training efficiency, the DDPG algorithm introduces two target networks, namely the target policy network μ(s; θ - ) and the target value network Q(s, a; ω - ), their update method uses soft update, that is, by filtering the parameters of the main network, the detailed update formula is as follows.

[0085] θ - new ← τ·θ new + (1-τ)θ - now

[0086] ω - new ← τ·ω new + (1-τ)ω - now

[0087] Where τ is the default parameter, because the learning process becomes relatively slower, more time is needed for training. Therefore, more powerful hardware facilities are needed to meet the requirements of training.

[0088] As an embodiment, the specific algorithm execution process is as follows:

[0089]

[0090] The observation state of the DDPG agent includes: a difference between the center-of-mass side slip angle and an expected value, a difference between the yaw rate and an expected value, an integral of the center-of-mass side slip angle error, an integral of the yaw rate error, a measured value of the center-of-mass side slip angle, and a measured value of the yaw rate.

[0091] The reward design of the DDPG agent includes: the smaller the center-of-mass side slip angle error and the yaw rate error, the greater the reward; a boundary is set, and a penalty is given when the boundary is exceeded.

[0092]

[0093] The yaw rate error is: The center-of-mass side slip angle error is:

[0094] In the formula, K = m(l f / k r -l r / k f ) / (l f +l r ) 2 , K is a stability factor, which is an important parameter of the steady-state response of the vehicle. L = l f +l r , and μ is the road adhesion coefficient.

[0095] Training is performed in the Simulink / CarSim environment until the desired effect is achieved, the training is stopped, and the model parameters are saved.

[0096] Embodiment 2

[0097] An embodiment of the present disclosure provides a rear-wheel active steering control system based on model prediction and reinforcement learning, comprising:

[0098] A model construction module is configured to construct a two-degree-of-freedom vehicle single-track model and design a model predictive controller based on the two-degree-of-freedom vehicle single-track model.

[0099] A prediction control module is configured to input the front-wheel steering angle, the vehicle speed, the yaw rate, and the center-of-mass side slip angle of the vehicle into the model predictive controller, take the given reference trajectory as the control target, and roll the output of the model predictive controller to output the optimal rear-wheel steering control amount.

[0100] The active steering control module is used for inputting vehicle state information and an optimal rear wheel steering angle control amount into a deep reinforcement learning network, using a multi-layer fully connected neural network as an action network of an Actor in a deep deterministic policy gradient reinforcement learning algorithm, setting an observation state value and a reward given a desired value, and using the deep deterministic policy gradient reinforcement learning algorithm to track a centroid side slip angle and a yaw rate target, so that the yaw rate and the centroid side slip angle are as close to the desired value as possible, and optimal rear wheel steering control is achieved.

[0101] Embodiment 3

[0102] In an embodiment of the present disclosure, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the model prediction and reinforcement learning based rear wheel active steering control method.

[0103] Embodiment 4

[0104] In an embodiment of the present disclosure, a non-transitory computer readable storage medium is provided, for storing computer instructions which, when executed by a processor, implement the model prediction and reinforcement learning based rear wheel active steering control method.

[0105] Embodiment 5

[0106] In an embodiment of the present disclosure, an electronic device is provided, comprising a processor, a memory, and a computer program; wherein the processor is connected with the memory, and the computer program is stored in the memory; when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device implements the model prediction and reinforcement learning based rear wheel active steering control method.

[0107] The present disclosure is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present disclosure. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices generate a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 The means for implementing one or more of the flows and / or blocks Figure 1 The means for implementing one or more of the flows and / or blocks

[0108] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable data processing devices to generate a computer implemented process, so that the instructions executed on the computer or other programmable data processing devices provide a process for implementing the functions specified in the flowchart Figure 1 one flow or multiple flows and / or the functions specified in the block Figure 1 one flow or multiple flows and / or the functions specified in the block

[0109] Although the specific embodiments of the present disclosure are described above with reference to the drawings, the description is not a limitation on the scope of protection of the present disclosure, and those skilled in the art should understand that various modifications or changes made on the basis of the technical solutions of the present disclosure without creative labor are still within the scope of protection of the present disclosure.

Claims

1. A rear wheel active steering control method based on model prediction and reinforcement learning, characterized in that, The application relates to a vehicle active steering control method based on model predictive control and deep reinforcement learning. The application comprises the following steps: A two-degree-of-freedom vehicle single-track model is constructed, and a model predictive controller is designed based on the two-degree-of-freedom vehicle single-track model. The front wheel steering angle, the vehicle speed, the yaw rate and the centroid side slip angle of the vehicle are input into the model predictive controller, a given reference trajectory is taken as a control target, the output of the model predictive controller is rolled and optimized, and the optimal rear wheel steering angle control amount is obtained. 2.The model prediction and reinforcement learning based rear wheel active steering control method of claim 1, wherein, The vehicle state information and the optimal rear wheel steering angle control amount are input into a deep reinforcement learning network, a multilayer fully connected neural network is taken as an action network of an actor in a deep deterministic policy gradient reinforcement learning algorithm, an expected value is given, an observation state value and a reward are set, the centroid side slip angle and the yaw rate target tracking are performed by using the deep deterministic policy gradient reinforcement learning algorithm, the yaw rate and the centroid side slip angle are made to be as close to the expected value as possible, and optimal rear wheel steering control is realized. Take ; A two-degree-of-freedom vehicle single-track model is constructed, and the model predictive controller is designed based on the two-degree-of-freedom vehicle single-track model. The dynamic model matrix is obtained as shown in the following formula: wherein, k f 、k r respectively the front and rear cornering stiffnesses; L f , L r is the distance from the front axle / rear axle to the center of mass, m is the total vehicle mass, v is the vehicle longitudinal speed, I z is the total vehicle moment of inertia about the z axis, is the center of mass side slip angle, gamma is the yaw rate, is the front wheel steering angle, is the rear wheel steering angle. 3.The model prediction and reinforcement learning based rear wheel active steering control method of claim 1, wherein, A model predictive controller is designed based on a two-degree-of-freedom vehicle monorail model, including: discretizing a continuous equation to obtain a new discrete state space model, predicting a step time domain, and the prediction equation is: 。 4.The model prediction and reinforcement learning based rear wheel active steering control method of claim 1, wherein, In the formula, The reference amount is set to The target function and the corresponding control constraint condition constitute a quadratic programming problem, and the optimal rear wheel steering increment is obtained by solving the problem through a solver . 5.The model prediction and reinforcement learning based rear wheel active steering control method of claim 1, wherein, The front wheel steering angle, the vehicle speed, the yaw rate and the centroid side slip angle of the vehicle are input into the model predictive controller, a given reference trajectory is taken as a control target, the output of the model predictive controller is rolled and optimized, and the optimal rear wheel steering angle control amount is obtained. 6.The model prediction and reinforcement learning based rear wheel active steering control method of claim 5, wherein, The observation state of the actor in the deep deterministic policy gradient reinforcement learning algorithm includes the difference between the centroid side slip angle and the expected value, the difference between the yaw rate and the expected value, the centroid side slip angle error integral, the yaw rate error integral, the centroid side slip angle measurement value and the yaw rate measurement value; the reward is set to be larger when the centroid side slip angle error and the yaw rate error are smaller; and a boundary is set to give a penalty when the boundary is exceeded.

7. A rear wheel active steering control system based on model predictive and reinforcement learning, specifically implementing the rear wheel active steering control method based on model predictive and reinforcement learning as claimed in any one of claims 1-6, characterized in that, The deep deterministic policy gradient reinforcement learning algorithm is based on Actor-Critic, uses a policy network to make a decision output, uses a value network to score the decision executed by the policy network, uses experience replay and a target network, and uses a TD algorithm to make the prediction of the value network closer to a target and make the prediction value of the target network closer to a real value function. The application relates to a vehicle active steering control method based on model predictive control and deep reinforcement learning. The model predictive control module is used for inputting the front wheel steering angle, the vehicle speed, the yaw rate and the centroid side slip angle of the vehicle into the model predictive controller, taking a given reference trajectory as a control target, rolling and optimizing the output of the model predictive controller, and obtaining the optimal rear wheel steering angle control amount. The active steering control module is used for inputting the vehicle state information and the optimal rear wheel steering angle control amount into the deep reinforcement learning network, taking a multilayer fully connected neural network as an action network of an actor in a deep deterministic policy gradient reinforcement learning algorithm, giving an expected value, setting an observation state value and a reward, performing centroid side slip angle and yaw rate target tracking by using the deep deterministic policy gradient reinforcement learning algorithm, making the yaw rate and the centroid side slip angle as close to the expected value as possible, and realizing optimal rear wheel steering control.

8. A computer program product comprising a computer program, characterized in that, The computer program is executed by a processor to implement the model prediction and reinforcement learning based rear wheel active steering control method in any one of claims 1-6.

9. A non-transitory computer-readable storage medium, comprising: The non-transitory computer readable storage medium is used to store computer instructions, which are executed by a processor to implement the model prediction and reinforcement learning based rear wheel active steering control method in any one of claims 1-6.

10. An electronic device, comprising: Comprise: A processor, a memory and a computer program; wherein the processor is connected with the memory, and the computer program is stored in the memory; when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the model prediction and reinforcement learning based rear wheel active steering control method in any one of claims 1-6.

Images