Trajectory tracking control method and system for automatic driving vehicle

Abstract

The invention belongs to the technical field of automatic driving, and particularly relates to an automatic driving vehicle trajectory tracking control method and system which comprises an information acquisition module, a steering controller and a data processing module. The information acquisition module acquires the position, speed and course angle information of the vehicle; the steering controller sends the pose information of the vehicle to the controller; the controller uses a data processing module to calculate an expected steering angle according to a trajectory tracking algorithm and pose information at the current moment in combination with an expected trajectory and then sends the expected steering angle to the steering controller, the steering controller controls front wheels to steer through an electronic steering wheel, and meanwhile the steering controller monitors the steering angle of the front wheels in real time through a steering angle sensor. The method has the advantages that aiming at the trajectory tracking control of the self-driving automobile, the control strategy integrating learning and passive control is researched, the regulation factor is designed, and the high-precision trajectory tracking of the self-driving automobile is realized.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of automatic driving, and particularly relates to an automatic driving vehicle trajectory tracking control method and system. BACKGROUND

[0002] With the rapid development of automatic driving technology, automatic driving vehicles are gradually moving towards practical application in the global range, and have become a core component of intelligent transportation systems. However, how to achieve efficient and accurate path tracking is still an important problem in the development of current automatic driving technology.

[0003] Currently, the control algorithm for trajectory tracking is mainly divided into two categories. One is path tracking based on kinematic model, such as pure pursuit algorithm and Stanley method. The other is dynamic model design path tracking, which improves the accuracy of control by considering the dynamic characteristics of the vehicle, such as PID control, LQR control, sliding mode control, adaptive control, fuzzy control and model predictive control (MPC) and the like. The application of neural networks in vehicle lateral control has experienced gradual evolution from modeling assistance to deep involvement in control strategy design, showing strong nonlinear modeling capability and adaptive control potential. Although the existing automatic driving path tracking technology has made significant progress, model uncertainty, computational complexity, parameter tuning and adaptability, and external disturbance, etc. are still the core problems and challenges that exist universally. SUMMARY

[0004] Based on the above problems, the application provides an automatic driving vehicle trajectory tracking control method and system. Compared with traditional advanced methods (such as MPC and LQR) that rely on model accuracy and numerical optimization, the port controlled Hamilton control framework can naturally guarantee the stability and robustness of the system under uncertainty. This energy-based controller is particularly suitable for high-robustness trajectory tracking tasks. The combination of neural networks and port controlled Hamilton controller realizes a complementary balance between the strictness based on physics and the adaptability driven by data. The technical scheme is as follows: An automatic driving vehicle trajectory tracking control method, comprising the following steps: S1. establishing a vehicle dynamics model and establishing a lateral error dynamic equation; S2. converting the nonlinear dynamics model into a port interconnected Hamilton equation for autonomous vehicle control; S3. designing a control variable, and converting an open-loop Hamilton system into a closed-loop Hamilton system through a matching equation; S4. using a recursive high-order neural network model to optimize and solve the calculation of the front wheel steering angle; S5. using the adjustable weighted parameters of the recursive high-order neural network to eliminate the influence of the uncertain items introduced by the system model.

[0005] The lateral error dynamics equation is preferably: ; ; ; ; ; the tracking error of the yaw angle, the lateral deviation, is the yaw angle of the vehicle, the yaw rate, is the front wheel steering angle of the vehicle; the matrix , and are expressed as: ; ; ; and respectively represent the cornering stiffness of the front and rear wheels, m is the mass of the vehicle, and respectively represent the longitudinal and lateral velocities, represents the moment of inertia around the Z axis, and represent the distances from the center of mass to the front and rear axles.

[0006] The nonlinear dynamic model is preferably re-described as a port-controlled Hamiltonian form for autonomous vehicle control: ； wherein , and are described as follows: ; ; ; ; represents the Hamiltonian function of the open-loop system.

[0007] The Hamiltonian function of the open-loop system is preferably differentiated to obtain: ; Due to Therefore: ; The dissipation inequality is established: ; External input c, represents the state variable, is the output; if the Hamiltonian function is semi-positive definite, the inequality satisfies the passivity condition.

[0008] Preferably, the open-loop port-controlled Hamiltonian system is reconstructed into the desired closed-loop port Hamiltonian system structure form through the desired interconnection and damping matrices, and the control variable obtained by solving the matching equation converts the open-loop Hamiltonian form into the desired closed-loop Hamiltonian form, ensuring that the closed-loop system containing the controller still satisfies the port-controlled Hamiltonian system form: ; The structure of the desired matrix and is as follows: ; ; and represent the desired interconnection matrix and the damping matrix, respectively; , , , , , and are design parameters, ensure the positive definiteness of the damping distribution; The desired Hamiltonian function is defined as: .

[0009] Preferably, based on the vehicle dynamics model, the yaw rate is redefined as , and the state vector and the input vector of the neural network are defined as: ;

[0010] In the formula, and are the estimated states of the lateral speed and the yaw rate, is the front wheel steering angle of the vehicle, and based on the above equation, the recursive high-order neural network model of the vehicle can be derived as follows: ; The error of the lateral velocity at k can be calculated as: ; The error of the lateral velocity at k, The error of the yaw rate at k; The estimator based on is defined as: ; where, and are measurement noises and stochastic processes with variances and The trainable weight vector is iteratively optimized by an adaptive learning algorithm; ; The estimated weight is the weight vector estimate of the th neuron at iteration , which is used as the main adaptive parameter in the recurrent high-order neural network model; the adaptive gain controls the influence of the prediction error on the weight update and indicates the correction magnitude; the covariance matrix reflects the uncertainty of the weight estimate and the confidence of the network learning; the regularization θ stabilizes numerical computation and mitigates divergence during matrix inversion; the sensitivity vector comes from the activation function of the recurrent high-order neural network , which captures the sensitivity of the output to input perturbations; the noise covariance is a scaled identity matrix that simulates the effect of observation noise; the covariance adjustment is a positive semi-definite matrix that prevents premature convergence; the learning rate controls the step size of the weight update.

[0011] Preferably, the steady-state yaw rate is derived from the design trajectory based on the curvature of the current position and the instantaneous longitudinal velocity: ; where, denotes the path curvature, denotes the instantaneous longitudinal velocity of the vehicle, denotes the desired steady-state yaw rate; By introducing a constraint condition for the front wheel steering angle: ； Using a solver, the following solution is obtained: ; The optimal value of the front wheel steering angle is finally solved by solving . .

[0012] Preferably, the recursive high-order neural network compensation term offsets the influence of the uncertainty of the model through adjustable weighting parameters , and the formula is as follows: ; Wherein is the , adjust the contribution between the Hamiltonian control framework and the neural network compensation to offset the influence of the uncertainty of the model.

[0013] An automatic driving vehicle trajectory tracking control system comprises an information acquisition module, a steering controller and a data processing module. The information acquisition module acquires the position, speed and heading angle information of the vehicle; the steering controller uses the data processing module to calculate the expected steering angle according to the trajectory tracking algorithm, the current pose information and the expected trajectory, and then sends the expected steering angle to the steering controller; the steering controller drives the electronic steering wheel through the output control quantity to control the steering of the front wheel, and at the same time, the steering controller monitors the steering angle of the front wheel in real time through the steering angle sensor to form a closed-loop feedback control.

[0014] Preferably, the data processing module comprises a vehicle lateral dynamics model, a port-controlled Hamiltonian controller, a recursive high-order neural network and a steering angle solver; and the learning and the passive controller weighting parameter design are fused.

[0015] Compared with the prior art, the application has the following beneficial effects:

[0016] Compared with the traditional advanced methods (such as MPC and LQR) which depend on model accuracy and numerical optimization, the port-controlled Hamiltonian controller improves the stability and robustness of the system under uncertainty. The combination of neural networks and port-controlled Hamiltonian systems achieves a complementary balance between the strictness based on physics and the adaptability driven by data. The port-controlled Hamiltonian control framework guarantees the stability of the controller through energy conservation, interconnection and dissipation structure, thereby ensuring the physical characteristics of the system. The neural network has strong nonlinear approximation and adaptive ability, which can compensate for unmodeled dynamics, parameter uncertainty and environmental changes. After embedding the neural network into the port-controlled Hamiltonian control framework, the overall controller can not only maintain the explainability and stability of the energy-based modeling, but also improve the adaptability and robustness in complex and safety-critical trajectory tracking scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The vehicle trajectory tracking control system; Figure 2 A diagram of a vehicle motion decomposition model; Figure 3 A diagram of a neural network architecture; Figure 4 A diagram of the overall structure of a controller; Figure 5 A diagram of the path tracking performance of a controller under a closed trajectory; Figure 6 A diagram of the path tracking performance of a controller under a specific turning path. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0019] The present application proposes an automatic driving vehicle trajectory tracking control method to solve the problem of trajectory tracking of an automatic driving vehicle, so as to realize high-precision trajectory tracking of the automatic driving vehicle. Compared with traditional advanced methods (such as MPC and LQR) which depend on model accuracy and numerical optimization, the port-controlled Hamilton control framework improves the stability and robustness of the system under uncertainty. The neural network is embedded in the port-controlled Hamilton control framework, achieving a complementary balance between the strictness based on physics and the adaptability driven by data. The port-controlled Hamilton control framework has inherent passivity, which guarantees stability through energy conservation, interconnection and dissipation structure, thereby ensuring the physical consistency of control. The neural network has strong nonlinear approximation and adaptive ability, which can compensate for unmodeled dynamics, parameter uncertainty and environmental changes. The controller can not only maintain the explainability and stability based on energy-based modeling, but also improve the adaptability and robustness in complex and safety-critical trajectory tracking scenarios. The method comprises: overall design of a vehicle trajectory tracking control system; establishment of a vehicle lateral dynamics model; design of a port-controlled Hamilton controller; design of a recursive high-order neural network steering controller; design of a steering angle solver; and design of a learning and passivity-based controller weighting parameter.

[0020] Step 1: Overall design of a vehicle trajectory tracking control system: The vehicle trajectory tracking control system design is as follows Figure 1As shown, the dual-antenna RTK-GNSS system collects the position, speed and heading angle information of the vehicle, and the steering controller sends the vehicle's pose information to the controller. The controller calculates the expected steering angle according to the trajectory tracking algorithm, the current pose information and the expected trajectory, and sends it to the steering controller. The steering controller controls the front wheel steering through the electronic steering wheel, and at the same time, the steering controller monitors the steering angle of the front wheel in real time through the steering angle sensor, forming a closed-loop feedback control.

[0021] Step 2: Establish the vehicle dynamics model and lateral error dynamic equation: The model is based on the following assumptions: the vehicle is a rigid body, the mass is evenly distributed on both sides; ignore the vertical motion of the vehicle, only consider its motion in a two-dimensional plane; ignore the influence of the vehicle suspension; assume that the rotation angle and rotation speed of the left and right wheels of the vehicle are the same. According to the vehicle model shown in the figure, the lateral dynamics equation of the vehicle is established as Figure 2 where represents the mass of the vehicle, and represent the longitudinal and lateral speeds respectively, represents the moment of inertia around the Z axis, is the yaw angle of the vehicle. The total longitudinal force of the front and rear wheels is and , and the total lateral force of the front and rear wheels is and , and represent the distance from the center of mass to the front and rear axles, is the front wheel steering angle of the vehicle.

[0022] Assuming that the tire side slip angle is small, the lateral tire force is a function of the tire side slip angle, and its expression is: where and represent the side stiffness of the front and rear wheels, and represent the side slip angles of the front and rear wheels, and according to the motion state of the vehicle and the geometric relationship, the expressions of the front and rear wheel side slip angles can be derived as follows:

[0023] According to the geometric relationship shown in the figure, the relationship between the lateral deviation and the required lateral acceleration can be expressed as: ;​​​​ Desired lateral acceleration and desired yaw rate The relationship between them can be derived mathematically: .

[0024] Similarly, from the reference trajectory of the design, the path curvature at a given time can be determined. and longitudinal velocity To obtain the desired yaw rate : .

[0025] Based on the above formula, we can deduce that: ; in, This represents the tracking error of the yaw angle. The derivative can be obtained by integrating the above equation: ; in , is the integral of the tracking error of the yaw angle with respect to the longitudinal velocity. Similarly, The derivative can be expressed based on the geometric relationships during vehicle motion: ; By making these two expressions equal, we get: .

[0026] Based on the above equations, the dynamics of the transverse error can be expressed as: ; ; ; .

[0027] in Let be the system's error state matrix. Let be the desired state matrix of the system. The system's control variable matrix, This refers to the steering angle of the vehicle's front wheels.

[0028] matrix , and It can be represented as: ; ; .

[0029] Step 3: Port-controlled Hamiltonian framework controller design: Port-controlled Hamiltonian systems emphasize the energy function, system interconnection structure, and dissipation mechanism, with good physical interpretability and system stability analysis capability. To adapt to the energy exchange and non-conservative characteristics in engineering systems, we use the interconnection and damping allocation passivity-based control (IDA-PBC) method of the port-controlled Hamiltonian framework to achieve the structured design of the path tracking control system. The port-controlled Hamiltonian system is a lumped parameter network model with independent energy storage elements, which can effectively represent the inherent energy exchange and dissipation dynamics in the system, and further develop the interconnection and dissipation allocation-passivity-based control (IDA-PBC). By restructuring the Hamiltonian structure and the desired energy function of the system, the control objective is transformed into an energy shaping problem, effectively improving the robustness and stability of the control system in the face of nonlinearities and external disturbances. We first describe the open-loop system as a standard form of the port-controlled Hamiltonian system: .

[0030] The equation set is the standard form of the port-controlled Hamiltonian system, where represents the state variables of the system, describing the energy state of the system, and is its time derivative, representing the dynamic evolution process of the system. The Hamiltonian function represents the total energy of the system. The matrix is the structural interconnection matrix, which is an anti-symmetric matrix satisfying , used to describe the internal conservative energy exchange structure of the system; is a symmetric positive semi-definite damping matrix, where , used to describe the energy dissipation process in the system. The matrix is the input distribution matrix, defining the way external input c acts on the system state; while the output usually represents the response variables of the system, used for feedback control design, reflecting the relationship between input and energy gradient. This modeling framework can systematically represent the storage, transmission, and dissipation characteristics of energy in mechatronic systems, providing a theoretical basis for passivity-based control design.

[0031] Autonomous vehicle systems inherently embody energy exchange dynamics, making the port-controlled Hamiltonian system modeling framework explicitly energy-centered, as its basis for suitable physical principles and stable controllers. In this work, the passivity-based method is strategically utilized to re-express the nonlinear dynamics as a port-controlled Hamiltonian form representation for autonomous vehicle control: ; where , and The following is described:

[0032] By differentiating the Hamiltonian function we obtain: Since we have: Thus, we establish the dissipation inequality:

[0033] If the Hamiltonian function is semi-positive definite, the inequality satisfies the passivity condition, thus proving the passivity of the system. Under this basis, a port-Hamiltonian (pH) system can be systematically reconstructed into a closed-loop port-controlled Hamiltonian system by interconnection and damping assignment, which still satisfies the Hamiltonian system structure form: Within the port-Hamiltonian framework, and denote the desired interconnection matrix and damping matrix, respectively. These matrices are key to the controller synthesis. The skew-symmetric matrix controls the conservative energy redistribution among the subsystems, thus preserving the total energy. In contrast, the symmetric semi-positive definite dissipation matrix captures the non-conservative energy loss, and the passivity-based approach aims to reconstruct the original system into the target closed-loop port-Hamiltonian form by strategically specifying , and the required Hamiltonian function , which is achieved by solving the partial differential matching conditions to compute the control law that enforces the desired energy and interconnection structure.

[0034] Based on the port-controlled Hamiltonian principle and the vehicle error dynamics model, the open-loop Hamiltonian function of the vehicle is defined as:

[0035] According to the target closed-loop port-Hamiltonian (pH) system described in the above equation, , and​​​​​​​​​​ The role of the definition of the equilibrium point , which corresponds to the isolated local minimum of , ensures that the system reaches asymptotic stability at

[0036] The passive controller, denoted as , from the port-Hamiltonian (pH) framework, ensures the asymptotic stability of the closed-loop system, defined as follows: ; where , , , , , and are design parameters, ensuring the positive definiteness of the damping distribution. The corresponding specific parameters are given in Table 1 below: Table 1 Specific parameters .

[0037] The desired Hamiltonian function is defined as: ; The structure of the desired matrices and is as follows: ; .

[0038] These matrices satisfy the structural conditions of the interconnection matrix and the damping matrix within the Hamiltonian framework. Therefore, the original Hamiltonian system can be transformed into the desired corresponding system by the control variables derived from the vehicle error dynamics model. The matching equation is represented as: .

[0039] Step 4: Recursive High-Order Neural Network Steering Controller Design: Recursive High-Order Neural Network (RHONN) is an architecture that integrates time memory and nonlinear modeling functions, making it particularly effective in representing complex, dynamic, and coupled nonlinear systems. The Recursive High-Order Neural Network (RHONN) model is generally represented as: ; where is the system state at time step , denotes the external input vector, is the high-order activation vector. This vector contains the nonlinear combination of the current state and input, such as polynomial expansion, cross terms or nonlinear basis functions. The structure of the network is shown in Figure 3 The trainable weight vector is updated by gradient descent, enabling the network to capture the nonlinear dynamics and time dependence within the system. The output represents the predicted state at the next time step, which is used for control input generation or state estimation. Each element forms a smooth vector of high-order polynomials, defined as: Let denote an unordered subset of the index set , for each , the coefficient is a non-negative integer, where the first indices are the coordinates of the Euclidean space . is defined as follows:

[0040] where S is the hyperbolic tangent activation function: ; The parameter serves as an output scaling factor to adjust the amplitude of the activation, while modulates the slope of the input transformation, thereby determining the degree of nonlinearity, both parameters are set to 0.25.

[0041] Based on the vehicle dynamics model, the yaw rate is redefined as , and then the state vector and input vector of the neural network are defined as: ; where and are the estimated states of RHONN, based on the above equation, the RHONN model of the vehicle can be derived as follows: .

[0042] The error of .

[0043] Based on the estimator , the definition is: ;

[0044] where and is a random process measuring noise and having variance and The trainable weight vector is iteratively optimized by the H∞ adaptive learning algorithm, enabling accurate modeling of the system's nonlinear dynamics and thus enhancing the robustness and convergence of the control policy.

[0045] .

[0046] Within the weight update framework based on the variables are defined as follows: the estimated weight is the weight vector estimate of the th neuron at iteration , which serves as the main adaptive parameter in the recurrent high-order neural network model, as it varies differently with iterations, initialized as a random numerical matrix of size 7 rows by 1 column; the adaptive gain controls the influence of the prediction error on the weight update, indicating the correction magnitude. The covariance matrix reflects the uncertainty of the weight estimates and the confidence of the network learning, initialized as a unit matrix scaled by 0.1. The regularization θ stabilizes numerical computation and mitigates divergence during matrix inversion, set to 0.01. The sensitivity vector comes from the activation function of the recurrent high-order neural network, capturing the sensitivity of the output to input perturbations. The noise covariance is usually a unit matrix scaled by 0.5 to simulate the effect of observation noise. The covariance adjustment is a positive semi-definite matrix to prevent premature convergence, set as a unit matrix scaled by 0.5, and the learning rate controls the weight update step size, set to 0.1, balancing the convergence speed and accuracy.

[0047] Step 5: Design of the steering angle solver: The steady-state yaw rate is derived from the design trajectory based on the curvature of the current position and the instantaneous longitudinal speed: ; where denotes the path curvature, denotes the instantaneous longitudinal speed of the vehicle, represents the desired steady-state yaw rate. Lateral velocity and yaw rate measurements can be obtained through simulation, while the target trajectory provides a reference steady-state yaw rate. Using these inputs, the front-wheel steering angle is computed. Traditional linear models fail to capture the complex system dynamics adequately, while the recurrent high-order neural network model utilizes high-order polynomial terms to approximate the nonlinear behavior, enabling accurate modeling of vehicle dynamics and improving prediction accuracy. Furthermore, by introducing a constraint on the front-wheel steering angle, the problem is formulated as a constrained nonlinear optimization problem: ; subjectto: .

[0048] The front-wheel steering angle is constrained to be within the range of 0 rad to 0.2 rad to satisfy vehicle design limitations. The optimization problem is implemented using a solver, resulting in the following solution: ; The optimal value of the front-wheel steering angle is found by solving .

[0049] Step 6: Fusing learned passivity controller weighting parameter design: The designed front-wheel steering angle consists of two components: one is derived from the controller based on the vehicle error dynamics model using the port-Hamiltonian method, and the other is a compensation term generated by the recurrent high-order neural network (RHONN) model through constrained optimization. To minimize tracking errors, the RHONN compensation term counteracts the model's uncertainties through an adjustable weighting parameter α, and the combined controller is redesigned as ; where adjusts the contribution between the Hamiltonian control framework and the neural network compensation. Utilizing its high-order polynomial structure, the RHONN captures system dynamics and nonlinear relationships to enhance generalization capabilities. However, residual errors may arise due to unmodeled effects. Optimizing can mitigate such errors while improving the controller's stability.

[0050] Experimental evaluation of the system is conducted by assessing path tracking accuracy, stability margins, and robustness under various operating conditions, which will make it more accurate. This analysis determines the optimal balance of performance trade-offs, establishing it as a key controller parameter.

[0051] Table 1 Parameter Table

[0052] Table 3 Parameter table two

[0053] Table 4 Parameter table three

[0054] An automatic driving vehicle trajectory tracking control system, comprising an information acquisition module, a steering controller and a data processing module; The information acquisition module acquires the position, speed and heading angle information of the vehicle; the steering controller uses the data processing module to calculate the expected steering angle according to the trajectory tracking algorithm, the current pose information and the expected trajectory, and sends it to the steering controller; the steering controller controls the front wheel steering through the electronic steering wheel, and at the same time the steering controller monitors the steering angle of the front wheel through the steering angle sensor in real time, forming a closed loop feedback control.

[0055] The data processing module includes a vehicle lateral dynamics model, a port-controlled Hamilton module, a recursive high-order neural network and a steering angle solver; the controller weighting parameter design is based on learning and passivity control; the port-controlled Hamilton module has inherent passivity, and the recursive high-order neural network adjustable weighting parameter is used to offset the uncertainty of the model.

[0056] In order to evaluate the effectiveness of the learning and passivity (LI-PBC) controller, a joint simulation experiment is carried out to verify the performance of the proposed controller. First, the optimal range of parameters under various working conditions is determined by iterative tuning of the system , the path tracking accuracy and system stability under discrete values are analyzed to determine an optimal value that can achieve balanced performance as the key parameter of the controller, and finally the value of is determined to be 0.85 for optimal effect. Subsequently, in order to verify the control effect of the proposed learning and passivity controller (LI-PBC) strategy, first, the predefined reference path is used to evaluate the closed path tracking performance. Figure 5 and Figure 6 show the ability of the controller to closely follow the reference path. In the proposed control framework, the interconnection and damping assignment passivity (IDA-PBC) controller first calculates the front wheel steering angle based on the error dynamics model. Then, RHONN (recursive high-order neural network) is used to represent the nonlinear dynamic characteristics of the vehicle. The solver calculates the compensated front wheel steering angle at each time step. The parameter is used to adjust the contribution between the Hamilton control framework and the neural network compensation, offset the uncertainty of the model, and ensure the control accuracy and real-time performance.

Claims

1. A trajectory tracking control method for an autonomous vehicle, characterized by, The method comprises the following steps: S1. Establishing a vehicle dynamics model and establishing a lateral error dynamic equation; S2. Converting the nonlinear dynamics model into a port interconnected Hamilton equation for autonomous vehicle control; S3. Designing a control variable to convert an open-loop Hamilton system into a closed-loop Hamilton system through a matching equation; S4. Using a recursive high-order neural network model to optimize the calculation of the front wheel steering angle; S5. Using the adjustable weighting parameters of the recursive high-order neural network to eliminate the influence of uncertain items introduced by the system model.

2. The trajectory tracking control method for an automated vehicle according to claim 1, wherein The lateral error dynamic equation is: ； ； ； ； ； tracking error representing a yaw angle, lateral deviation, a yaw angle of the vehicle, yaw rate, a front wheel steering angle of the vehicle; matrix , and is represented as: ； ； ； and Kf and Kr represent the cornering stiffness of the front and rear wheels, respectively, m is the mass of the vehicle, and Vx and Vy represent the longitudinal and lateral velocities, respectively, Iz represents the moment of inertia about the Z axis, and L and Lr represent the distance from the center of mass to the front and rear axles, respectively.

3. The trajectory tracking control method for an automated vehicle according to claim 1, wherein, The nonlinear dynamics model expression is converted into a structured port-Hamilton pH equation port interconnected Hamilton equation for autonomous vehicle control: ； wherein , and are described as follows: ； ； ； ； represents the Hamiltonian function.

4. The trajectory tracking control method for an automated vehicle according to claim 3, wherein A control variable is designed to convert an open-loop Hamilton system into a closed-loop Hamilton system through a matching equation: ； Desired matrix and The structure of the following: ； ; and denote the desired interconnection matrix and the damping matrix, respectively; , , , , , and are design parameters, ensuring positive definiteness of the damping assignment; Desired Hamiltonian is defined as: 。 5. The trajectory tracking control method for an autonomous vehicle according to claim 1, wherein, Based on the vehicle dynamics model, the yaw angular velocity is redefined as The state vector of the neural network and the input vector are then defined as: ； ; wherein and is the lateral velocity, the yaw rate estimation state, is the front wheel steering angle of the vehicle, based on the above equations, the recursive high-order neural network model of the vehicle can be derived as follows: ； The error is obtained through calculation: ; is the error in lateral velocity at time k, is the error in yaw rate at time k. Based on the estimator is defined as: ； wherein and are random processes with variance and The trainable weight vector is iteratively optimized by a H∞ adaptive learning algorithm. ； estimated weight is the iteration the weight vector estimate of the th neuron used as the main adaptive parameter in the recurrent higher-order neural network model; adaptive gain controls the influence of the prediction error on the weight update, indicating the correction magnitude; covariance matrix reflects the uncertainty of the weight estimate and the confidence of the network learning; regularization theta stabilizes numerical computation and mitigates divergence during matrix inversion; sensitivity vector activation function from the recurrent higher-order neural network , capturing the sensitivity of the output to input perturbations; noise covariance is a scaled identity matrix simulating the effect of observation noise; covariance adjustment is a semi-positive definite matrix that prevents premature convergence; learning rate controls the weight update step size.

6. The automatic driving vehicle trajectory tracking control method according to claim 5, characterized in that, The steady-state yaw rate is derived from the designed trajectory according to the curvature of the current position and the instantaneous longitudinal speed: ； wherein denotes the path curvature, denotes the instantaneous longitudinal speed of the vehicle, denotes the desired steady-state yaw rate; By introducing the constraint condition of the front wheel steering angle: ； ； The following solution is obtained using a solver: ； The optimal value of the front wheel steering angle is obtained by solving .​ 7. The trajectory tracking control method for an automated vehicle according to claim 1, wherein Recursive higher order neural network compensation terms with adjustable weighting parameters The uncertainty of the model is counteracted by the formula: ； wherein are calculated in the Hamiltonian model , Adjust the contribution between the Hamiltonian control framework and the neural network compensation to offset the influence of model uncertainty.

8. An automatic driving vehicle trajectory tracking control system employing the method of any one of claims 1-7. The system comprises an information acquisition module, a steering controller, and a data processing module. The information acquisition module acquires the position, speed, and heading angle information of the vehicle; the steering controller uses the data processing module to calculate the expected steering angle according to the trajectory tracking algorithm, the current pose information, and the expected trajectory, and sends the expected steering angle to the steering controller; the steering controller controls the front wheel steering through the electronic steering wheel, and simultaneously monitors the steering angle of the front wheel in real time through the steering angle sensor to form a closed-loop feedback control.

9. The automatic trajectory tracking control system for a vehicle according to claim 8, wherein The data processing module comprises a vehicle lateral dynamics model, a port controlled Hamilton module, a recursive high-order neural network, a steering angle solver, a learning-based and passivity control controller weighting parameter design, and an internally passive port controlled Hamilton module that uses the adjustable weighting parameters of the recursive high-order neural network to offset the uncertainty of the model.

Images