An automatic driving vehicle path tracking control method and device

Abstract

The application belongs to the technical field of control, and discloses an automatic driving vehicle path tracking control method and device. The method is characterized in precise representation of vehicle movement and dynamic characteristics by constructing a nonlinear path tracking model and a dynamic model; a comprehensive path tracking error dynamic model is constructed to convert path tracking into an error convergence problem and simplify the control target; a nonlinear extended state observer is constructed to effectively observe unobservable states, nonlinear dynamics and uncertain disturbances of the system; a front wheel rotation angle control amount is calculated based on real-time estimated values obtained through observation, which can compensate for control deviation caused by unknown dynamics and disturbances; and finally, the vehicle is controlled according to the rotation angle control amount, which can significantly improve path tracking accuracy and robustness within a wide vehicle speed range, and realize stable and accurate automatic driving path tracking control.

Description

Technical Field

[0001] This application relates to the field of control technology, and in particular to a method and apparatus for path tracking control of autonomous vehicles. Background Technology

[0002] Path tracking, as a core component of motion control, directly determines the vehicle's driving accuracy and safety. Current vehicle chassis execution systems exhibit strong nonlinearity and motion coupling characteristics. During vehicle operation, they are susceptible to uncertain external disturbances such as crosswinds and uneven road surfaces. Furthermore, vehicle parameters exhibit time-varying characteristics under complex conditions such as high curvature. Traditional path tracking control methods often ignore or simplify system nonlinearity, model uncertainty, and external disturbances, resulting in poor path tracking accuracy and robustness across a wide speed range. This makes it difficult to meet the high-performance and high-stability path tracking control requirements of autonomous vehicles. Summary of the Invention

[0003] This application provides a path tracking control method and apparatus for autonomous vehicles, which can improve path tracking accuracy and robustness over a wide speed range, and achieve stable and accurate path tracking control for autonomous vehicles.

[0004] In a first aspect, this application provides a path tracking control method for an autonomous vehicle, the method comprising the following steps: Based on the vehicle's planar motion and dynamic characteristics, a nonlinear path tracking model and a dynamic model are obtained; Based on the nonlinear path tracking model and the dynamic model, a dynamic model of the comprehensive path tracking error, which is formed by the combination of lateral error and heading angle error with the aiming distance, is constructed; wherein, the dynamic model takes the comprehensive path tracking error as the system output and the front wheel steering angle control quantity as the system input; Based on the system output of the dynamic model, the measurable state variables of the vehicle, the reference path information of the vehicle, and the front wheel steering angle control quantity of the vehicle, a nonlinear extended state observer is constructed. The nonlinear extended state observer is used to obtain real-time estimates of unmeasurable states, unmodeled nonlinear dynamics, and uncertain disturbances in the dynamic model. Based on the real-time estimated value, the front wheel steering angle control amount of the vehicle is determined; The vehicle is controlled based on the front wheel steering angle control value.

[0005] Secondly, this application provides an autonomous vehicle path tracking control device, the device comprising: The first unit is used to obtain the nonlinear path tracking model and the dynamic model based on the vehicle's planar motion and dynamic characteristics; The second unit is used to construct a dynamic model of the comprehensive path tracking error formed by the combination of lateral error and heading angle error and the aiming distance, based on the nonlinear path tracking model and the dynamic model; wherein, the dynamic model takes the comprehensive path tracking error as the system output and the front wheel steering angle control quantity as the system input. The third unit is used to construct a nonlinear extended state observer based on the system output of the dynamic model, the measurable state variables of the vehicle, the reference path information of the vehicle, and the front wheel steering angle control quantity of the vehicle. The fourth unit is used to obtain real-time estimates of unmeasurable states, unmodeled nonlinear dynamics, and uncertain disturbances in the dynamic model using the nonlinear extended state observer. The fifth unit is used to determine the front wheel steering angle control amount of the vehicle based on the real-time estimated value; The sixth unit is used to control the vehicle based on the front wheel steering angle control amount.

[0006] As can be seen from the above technical solution, this application has the following beneficial effects compared with the prior art: This autonomous vehicle path tracking control method derives a nonlinear path tracking model and a dynamic model based on the vehicle's planar motion and dynamic characteristics, accurately characterizing the vehicle's motion and dynamics during path tracking. Then, based on the nonlinear path tracking model and the dynamic model, a dynamic model is constructed to represent the comprehensive path tracking error, combining lateral error, heading angle error, and aiming distance. This transforms the path tracking problem into a comprehensive error convergence control problem, simplifying the control objective. Next, a nonlinear extended state observer is constructed based on the system output of the dynamic model, the vehicle's measurable state variables, reference path information, and the front wheel steering angle control quantity. This enables effective observation of unmeasurable states, unmodeled nonlinear dynamics, and uncertain disturbances in the dynamic model. The nonlinear extended state observer is used to obtain real-time estimates of these unmeasurable states, unmodeled nonlinear dynamics, and uncertain disturbances. The front wheel steering angle control quantity is determined based on these real-time estimates, compensating for control deviations based on unknown information. Finally, the vehicle is controlled based on the front wheel steering angle control quantity, improving path tracking accuracy and robustness over a wide speed range and achieving stable and accurate autonomous vehicle path tracking control.

[0007] The further effects of the aforementioned non-conventional preferred method will be explained below in conjunction with specific embodiments. Attached Figure Description

[0008] To more clearly illustrate the embodiments of this application or the existing technical solutions, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 A flowchart illustrating an autonomous vehicle path tracking control method provided in this application; Figure 2 This is a schematic diagram of an intelligent vehicle path tracking system model; Figure 3 This is a schematic diagram of a path tracking controller based on active disturbance rejection control. Figure 4a A comparison of path tracking simulation results at a vehicle speed of 20 km / h between pure tracking control, LQR control, and the active disturbance rejection control proposed in this invention; Figure 4b A comparison of lateral error simulation results at a vehicle speed of 20 km / h between pure tracking control, LQR control, and the active disturbance rejection control proposed in this invention; Figure 4c A comparison of path tracking simulation results at a vehicle speed of 40 km / h between pure tracking control, LQR control, and the active disturbance rejection control proposed in this invention; Figure 4d A comparison of lateral error simulation results at a vehicle speed of 40 km / h between pure tracking control, LQR control, and the active disturbance rejection control proposed in this invention; Figure 4e A comparison of path tracking simulation results at a vehicle speed of 60 km / h between pure tracking control, LQR control, and the active disturbance rejection control proposed in this invention; Figure 4f A comparison of lateral error simulation results at a vehicle speed of 60 km / h between pure tracking control, LQR control, and the active disturbance rejection control proposed in this invention; Figure 5 A schematic diagram of the structure of an autonomous vehicle path tracking control device provided in this application; Figure 6 This is a schematic diagram of the structure of an electronic device provided in this application. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0011] The various non-limiting embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0012] The intelligent vehicle path tracking system model of this application is as follows: Figure 2 As shown. Specifically, as... Figure 1 As shown, in this embodiment, the method may include, for example, the following steps: S101: Based on the vehicle's planar motion and dynamic characteristics, a nonlinear path tracking model and a dynamic model are obtained.

[0013] Optionally, the process of obtaining the nonlinear path tracking model and dynamic model based on the vehicle's planar motion and dynamic characteristics includes: A two-degree-of-freedom vehicle dynamics model is established with the vehicle's center of gravity sideslip angle and yaw rate as state variables. This two-degree-of-freedom vehicle dynamics model is used as the dynamic model. The dynamic model is used to describe the dynamic relationship between the vehicle's lateral motion and yaw motion. The dynamic model includes unmodeled dynamic terms in the vehicle's lateral dynamics and yaw dynamics caused by model coupling nonlinearity, disturbances, and parameter uncertainties, as well as unmodeled dynamic terms in the front axle tire force and rear axle tire force. By using onboard sensors to acquire the vehicle's longitudinal velocity, lateral velocity, and heading angle in real time, and combining this with reference path information, a nonlinear path tracking model is constructed to describe the relationship between lateral error and heading angle error and the vehicle's motion.

[0014] It should be noted that this step forms the model foundation of the entire path tracking control method. Its purpose is to accurately describe the vehicle's trajectory and dynamic force relationships on the road surface using a mathematical model, providing a basis for subsequent error calculation, observer design, and controller design. In autonomous driving systems, without accurate kinematic and dynamic models, path tracking will suffer from problems such as large deviations, easy divergence, and high-speed instability. This step is a prerequisite for achieving high-precision control. The established models are all based on the actual physical characteristics of the vehicle, without excessive simplification, and can realistically reflect the nonlinearity, coupling, and uncertainty characteristics of the vehicle.

[0015] In this embodiment, a two-degree-of-freedom vehicle dynamics model can be established first: First, a two-degree-of-freedom vehicle dynamics model is established with the sideslip angle and yaw rate as state variables, and this model is directly used as the dynamics model in this invention.

[0016] Two-degree-of-freedom vehicle dynamics model: refers to a simplified dynamics model that only considers the lateral and yaw motion of the vehicle. It is the most commonly used and most engineering-friendly model form in autonomous driving path tracking control. Side slip angle: refers to the angle between the direction of the vehicle's center of gravity velocity and the longitudinal axis of the vehicle body, reflecting whether the vehicle is sideslipping; Yaw rate: refers to the angular velocity of a vehicle rotating around the Z-axis perpendicular to the ground, reflecting the speed and stability of the vehicle's steering. Unmodeled dynamic terms: These refer to nonlinearities, disturbances, parameter changes, and other dynamics that cannot be fully and accurately expressed by theoretical models and actually exist. They are the core objects that this invention needs to observe and compensate for.

[0017] This dynamic model is specifically designed to describe the coupled dynamic relationship between the lateral and yaw motions of a vehicle. It also fully includes two types of unmodeled dynamics: the first type consists of unmodeled dynamic terms in the lateral and yaw dynamics caused by model coupling nonlinearity, external disturbances, and parameter uncertainties; the second type consists of unmodeled dynamic terms in the front axle tire forces and rear axle tire forces, which characterize the nonlinear characteristics and uncertainties of the tires under different road surfaces, loads, and speeds.

[0018] Then, a non-linear path tracking model can be constructed: The vehicle uses onboard sensors to acquire longitudinal speed, lateral speed, and heading angle in real time, while also receiving reference path information from the autonomous driving decision-making and planning layer. Based on this, a nonlinear path tracking model is constructed.

[0019] Nonlinear path tracking model: A model based on vehicle planar kinematics that does not make linear approximations, which can maintain accuracy under large angle and large curvature conditions; Lateral error: The vertical distance from the vehicle's center of gravity to the center line of the reference path is the most critical error indicator in path tracking. Heading angle error: The difference between the actual heading of the vehicle and the tangent direction of the reference path, which directly affects the smoothness of tracking and tracking error.

[0020] The model's role is to describe in real time how lateral and heading angle errors change with vehicle movement, linking vehicle position, attitude, speed, and path deviation, and serving as the direct basis for constructing the comprehensive error later.

[0021] The nonlinear path tracking model is as follows:

[0022] in,

[0023] Dynamic model expression:

[0024] in,

[0025] S102: Based on the nonlinear path tracking model and the dynamic model, construct a dynamic model of the comprehensive path tracking error formed by the combination of lateral error and heading angle error with the aiming distance.

[0026] The dynamic model uses the comprehensive path tracking error as the system output and the front wheel steering angle control quantity as the system input.

[0027] Optionally, the step of constructing a dynamic model of the comprehensive path tracking error, which is formed by combining the lateral error and the heading angle error with the pre-aiming distance, based on the nonlinear path tracking model and the dynamic model, includes: The vehicle's current lateral error and current heading angle error are obtained, and the vehicle's current aiming distance is multiplied by the heading angle error and then added to the lateral error. The sum is used as the comprehensive path tracking error. The first and second derivatives of the time of the integrated path tracking error are obtained. In the process of differentiation, the derivatives of the lateral error and the heading angle error are replaced by the geometric kinematic relationship in the nonlinear path tracking model, and the small angle case is approximated. Substitute the dynamic relationship in the dynamic model, with the front wheel steering angle as input and the yaw rate and the center of gravity sideslip angle as state variables, into the terms related to the vehicle state change in the second derivative expression, so that the second derivative is finally expressed as the sum of the lumped unmodeled nonlinear and uncertain dynamic terms, the observable state terms, and the coefficient terms of the control gain multiplied by the front wheel steering angle control quantity. A dynamic model in input-output form is obtained, with the front wheel steering angle control quantity as input and the comprehensive path tracking error as output. This dynamic model is a second-order system and includes the lumped unmodeled nonlinear and uncertain dynamic terms. The lumped unmodeled nonlinear and uncertain dynamic terms, the unmodeled dynamics caused by model coupling nonlinearity in vehicle lateral dynamics and yaw dynamics, external disturbances and parameter uncertainties, tire unmodeled dynamics, and the rate of change of the approximation error are also included.

[0028] Understandably, this step integrates the dispersed lateral and heading angle errors into a unified control objective, transforming the path tracking problem into a comprehensive path tracking error convergence problem. The dynamic model constructed in this step uses the comprehensive path tracking error as the system output and the front wheel steering angle control variable as the system input.

[0029] The comprehensive path tracking error is obtained by taking the vehicle's current lateral error and current heading angle error. The aiming distance is multiplied by the sine of the heading angle error, and then added to the lateral error. The sum is taken as the comprehensive path tracking error.

[0030] Differentiate the comprehensive error and replace the kinematic relationship: Calculate the first and second time derivatives of the comprehensive path tracking error in sequence; during the differentiation process, replace the derivatives of the lateral error and the heading angle error with the geometric kinematic relationship in the nonlinear path tracking model, and approximate the small angle case to obtain the derivative expression containing the approximate error.

[0031] Substituting the dynamic relationship into the standard form: Substituting the dynamic relationship in the dynamic model, with the front wheel steering angle as input and the yaw rate and the center of gravity sideslip angle as state variables, into the terms related to the vehicle state change in the second derivative expression; finally, the second derivative is organized into the form of the sum of the lumped unmodeled nonlinear and uncertain dynamics, the observable state terms, and the control gain coefficient terms multiplied by the front wheel steering angle control quantity.

[0032] A second-order input-output dynamic model is obtained: a second-order system dynamic model is formed with the front wheel steering angle control quantity as input and the comprehensive path tracking error as output; the model contains lumped unmodeled nonlinear and uncertain dynamic terms, which integrate the unmodeled dynamics caused by model coupling nonlinearity, external disturbance and parameter uncertainty in vehicle lateral dynamics and yaw dynamics, the unmodeled dynamics of tires, and the rate of change of small-angle approximation error.

[0033] The overall path tracking error is:

[0034] First derivative of the overall error:

[0035] S103: Construct a nonlinear extended state observer based on the system output of the dynamic model, the measurable state variables of the vehicle, the reference path information of the vehicle, and the front wheel steering angle control quantity of the vehicle.

[0036] Optionally, the step of constructing a nonlinear extended state observer based on the system output of the dynamic model, the measurable state variables of the vehicle, the reference path information of the vehicle, and the front wheel steering angle control quantity of the vehicle includes: The dynamic model of the integrated path tracking error is transformed into a second-order system through variable substitution. The first state variable is defined as the integrated path tracking error, the second state variable is defined as the first derivative of the integrated path tracking error, and the lumped unmodeled nonlinear and uncertain dynamic terms in the dynamic model are expanded into a third state variable. The third state variable is used as the total disturbance to form a third-order extended state system. The measured value of the integrated path tracking error is used as the system output. A nonlinear extended state observer is constructed based on the third-order extended state system. The observer takes the system output and the front wheel steering angle control quantity as inputs and introduces nonlinear functions and gain parameters to update the estimated values ​​of each state variable online.

[0037] Optionally, the observer updates its estimates of each state variable according to the following rules: The rate of change of the estimated value of the first state variable is equal to the sum of the estimated value of the second state variable and the first nonlinear term. The first nonlinear term is obtained by dividing the first gain parameter by the cube of the preset scaling parameter and then multiplying it by the value of the observation error after the first nonlinear transformation. The observation error is the difference between the system output and the estimated value of the first state variable. The rate of change of the estimated value of the second state variable is equal to the sum of the first and second time-varying terms, the estimated value of the third state variable, the product of the coefficient term and the front wheel steering angle control quantity, and the second nonlinear term, which can be calculated using the measurable state variables of the vehicle and the reference path information. The second nonlinear term is obtained by dividing the second gain parameter by the preset scaling parameter and then multiplying the observation error by the value after the second nonlinear transformation. The rate of change of the estimated value of the third state variable is equal to the third gain parameter multiplied by the observation error after the third nonlinear transformation; The first to third nonlinear transformations all adopt the following rules: the output after the input value is nonlinearly transformed is equal to the absolute value of the input value raised to a specified power multiplied by the sign function value of the input value, and the specified powers used in the first to third nonlinear transformations are different. The specified powers are calculated according to the corresponding rules through a common preset base parameter; the values ​​of the first to third gain parameters make the observer error dynamic matrix a Herwitz matrix.

[0038] This step involves designing a nonlinear extended state observer (hereinafter referred to as the observer) for real-time observation of the system's unmeasurable states and total disturbances. The observer structure is shown in the attached diagram. Figure 3 accomplish.

[0039] Constructing a third-order extended state system transforms the dynamic model of the integrated path tracking error into a standard second-order system through variable substitution; Definition: First state variable x1: Overall path tracking error; Second state variable x2: First derivative of the overall path tracking error; The third state variable x3: the lumped unmodeled nonlinearity and uncertain dynamics are taken as the total disturbance; thus forming a third-order extended state system.

[0040] Establish a nonlinear extended state observer: The measured value of the integrated path tracking error is used as the system output, and the system output and the front wheel steering angle control quantity are used as the observer input. A nonlinear extended state observer is constructed based on the third-order extended state system. The observer updates the estimated values ​​of each state variable online according to the preset nonlinear function and gain parameter.

[0041] The observer state update rule is as follows: The observer updates the estimates of the three state variables according to the following rules: The rate of change of the estimated value of the first state variable is equal to the estimated value of the second state variable plus the first nonlinear term. The first nonlinear term is obtained by dividing the first gain parameter by the cube of the preset scaling parameter and then multiplying it by the value of the observation error after the first nonlinear transformation. The observation error is the difference between the system output and the estimated value of the first state variable.

[0042] The rate of change of the estimated value of the second state variable is equal to the sum of the first time-varying term, the second time-varying term, the estimated value of the third state variable, the product of the coefficient term and the front wheel steering angle, and the second nonlinear term. The second nonlinear term is obtained by dividing the second gain parameter by the preset scaling parameter and then multiplying it by the value of the observation error after the second nonlinear transformation.

[0043] The rate of change of the estimated value of the third state variable is equal to the value of the third gain parameter multiplied by the observation error after the third nonlinear transformation.

[0044] The first to third nonlinear transformation rules are unified as follows: Output value = Absolute value of input raised to a specified power × Sign function of input value; the powers of the three transformations are different, and are calculated according to the same preset base parameter; the values ​​of the first to third gain parameters ensure that the observer error dynamic matrix is ​​a Hurwitz matrix, thus ensuring observer stability. The Hurwitz matrix is... , , , These are the first gain parameter, the second gain parameter, and the third gain parameter, respectively.

[0045] The formula for a third-order extended state system is as follows:

[0046] Nonlinear extended state observer:

[0047] S104: Use the nonlinear extended state observer to obtain real-time estimates of unmeasurable states, nonlinear dynamics, and uncertain disturbances in the dynamic model.

[0048] Optionally, obtaining real-time estimates of unmeasurable states, nonlinear dynamics, and uncertain disturbances in the dynamic model using the nonlinear extended state observer includes: In each control cycle, the first state variable estimate of the output of the nonlinear extended state observer is obtained as the real-time estimate of the integrated path tracking error. The second state variable estimate output by the nonlinear extended state observer is used as the real-time estimate of the unmeasurable state in the dynamic model, where the unmeasurable state is the first derivative of the real-time estimate of the integrated path tracking error. The estimated value of the third state variable, which is estimated and output in real time by the nonlinear extended state observer based on its preset update rule, is used as the real-time estimate of the unmodeled nonlinear dynamics and uncertain disturbances in the dynamic model. The unmodeled nonlinear dynamics and uncertain disturbances ensemble includes unmodeled dynamics, parameter uncertainties, and external disturbances.

[0049] This step extracts all the estimation information required for control from the observer output, providing a basis for subsequent calculation of the steering angle control quantity.

[0050] Within each control cycle, perform the following extraction operations: The estimated value of the first state variable output by the observer is used as the real-time estimate of the integrated path tracking error; The estimated value of the second state variable output by the observer is used as the real-time estimate of the unmeasurable state in the dynamic model; this unmeasurable state is the same concept as the first derivative of the integrated path tracking error. The estimated value of the third state variable output by the observer is used as the real-time estimate of the unmodeled nonlinear dynamics and uncertain disturbances in the dynamic model; the total disturbance set includes unmodeled dynamics, parameter uncertainties and external disturbances.

[0051] S105: Determine the front wheel steering angle control amount of the vehicle based on the real-time estimated value.

[0052] Optionally, determining the front wheel steering angle control value of the vehicle based on the real-time estimated value includes: Using the vehicle's longitudinal acceleration, lateral acceleration, yaw rate, heading angle error, longitudinal velocity, and reference path curvature information, the first time-varying feedforward term and the second time-varying feedforward term are calculated in real time. The front wheel steering angle control quantity is calculated based on the real-time estimate of the total disturbance output by the nonlinear extended state observer, the first time-varying feedforward term and the second time-varying feedforward term, and the preset feedback term related to the scaling parameter and the observer gain, in a manner that combines feedforward compensation, disturbance compensation and error feedback. The feedback term is composed of the measured value of the integrated path tracking error multiplied by a first feedback coefficient, plus the real-time estimated value of the rate of change of the integrated path tracking error multiplied by a second feedback coefficient. The first feedback coefficient and the second feedback coefficient are determined by the square of the scaling parameter and the product of the scaling parameter and the preset stability parameter, respectively.

[0053] Optionally, the expression for calculating the front wheel steering angle control quantity is as follows: take the negative of the sum of the first time-varying feedforward term and the second time-varying feedforward term, subtract the real-time estimate of the total disturbance, add the feedback term, and divide the result by the control gain coefficient related to the vehicle parameters and the aiming distance to obtain the front wheel steering angle control quantity; wherein, the control gain coefficient is determined by the front axle tire lateral stiffness, the distance from the vehicle center of gravity to the front axle, the vehicle's moment of inertia about the vertical axis, and the aiming distance.

[0054] This step, based on disturbance compensation and error feedback, calculates the final front wheel steering angle control value, which is the core link in achieving high-precision tracking.

[0055] Calculate the time-varying feedforward term: Utilize the vehicle's longitudinal acceleration, lateral acceleration, yaw rate, heading angle error, longitudinal velocity, and reference path curvature information to calculate the first and second time-varying feedforward terms in real time.

[0056] Calculation of front wheel steering angle control quantity: A combination of feedforward compensation, disturbance compensation, and error feedback is adopted. The front wheel steering angle control quantity is calculated based on the real-time estimate of the total disturbance, two time-varying feedforward terms, and a preset feedback term. The feedback term is composed of the comprehensive path tracking error measurement value multiplied by the first feedback coefficient, plus the comprehensive error change rate estimate multiplied by the second feedback coefficient. The two feedback coefficients are determined by the square of the scaling parameter and the product of the scaling parameter and the preset stability parameter, respectively.

[0057] The formula for calculating the control quantity is as follows: the sum of the first time-varying feedforward term and the second time-varying feedforward term is negative, the real-time estimate of the total disturbance is subtracted, and the feedback term is added. The result is divided by the control gain coefficient to obtain the front wheel steering angle control quantity. The control gain coefficient is determined by the front axle tire lateral stiffness, the distance from the center of gravity to the front axle, the moment of inertia about the Z-axis, and the aiming distance.

[0058] Front wheel steering angle control ratio formula:

[0059] S106: Control the vehicle according to the front wheel steering angle control amount.

[0060] Optionally, before the step of controlling the vehicle according to the front wheel steering angle control amount, the method further includes: The calculated front wheel steering angle control value is subjected to amplitude limiting processing. If the front wheel steering angle control value exceeds the preset maximum steering angle amplitude, the front wheel steering angle control value is set to the maximum steering angle amplitude. The rate of change of the front wheel steering angle control value after amplitude limiting is applied; If the time change rate of the front wheel steering angle control quantity after amplitude limiting exceeds the preset maximum change rate, then the change rate of the front wheel steering angle control quantity after amplitude limiting will be limited to the maximum change rate. The preset maximum steering angle amplitude and the preset maximum rate of change are preset based on the physical constraints of the vehicle steering system.

[0061] This step applies the calculated control quantity to the vehicle steering actuator to complete the path tracking closed-loop control.

[0062] Before the output control quantity is sent to the vehicle, a safety limiting process is added to protect the steering system and ensure ride smoothness: Amplitude limiting: If the front wheel steering angle control value exceeds the preset maximum steering angle amplitude, it will be limited to the maximum steering angle amplitude. Change rate limit: If the change rate of the control quantity over time exceeds the preset maximum change rate after the limit is applied, its change rate is limited to within the threshold. The above maximum steering angle amplitude and maximum change rate are preset according to the physical constraints of the vehicle steering system.

[0063] After the amplitude limit is completed, the final front wheel steering angle control value is input into the vehicle steering actuator to control the vehicle to travel along the reference path. At different vehicle speeds such as 20km / h, 40km / h, and 60km / h, this method can maintain a small lateral error, and the tracking performance is significantly better than pure tracking control and LQR control. The corresponding results can be found in the appendix. Figure 4a See attached diagram Figure 4f The simulation results.

[0064] This autonomous vehicle path tracking control method derives a nonlinear path tracking model and a dynamic model based on the vehicle's planar motion and dynamic characteristics, accurately characterizing the vehicle's motion and dynamics during path tracking. Then, based on the nonlinear path tracking model and the dynamic model, a dynamic model is constructed to represent the comprehensive path tracking error, combining lateral error, heading angle error, and aiming distance. This transforms the path tracking problem into a comprehensive error convergence control problem, simplifying the control objective. Next, a nonlinear extended state observer is built based on the system output of the dynamic model, the vehicle's measurable state variables, reference path information, and the front wheel steering angle control quantity. This enables effective observation of unmeasurable states, unmodeled nonlinear dynamics, and uncertain disturbances in the dynamic model. The nonlinear extended state observer is used to obtain real-time estimates of these unmeasurable states, unmodeled nonlinear dynamics, and uncertain disturbances. The front wheel steering angle control quantity is determined based on these real-time estimates, compensating for control deviations caused by unknown dynamics and disturbances. Finally, the vehicle is controlled based on the front wheel steering angle control quantity, improving path tracking accuracy and robustness over a wide speed range, achieving stable and accurate path tracking control for autonomous vehicles.

[0065] To better understand this application, specific embodiments are illustrated below.

[0066] Step 1: Based on the vehicle's planar motion, construct a nonlinear path tracking model for the vehicle; considering the vehicle's dynamic characteristics and its nonlinear and uncertain dynamics, establish a dynamic model for the vehicle.

[0067] The schematic diagram of the vehicle path tracking planar motion is as follows: Figure 2 As shown, based on the kinematics of autonomous vehicles, a nonlinear path tracking model of the following form is established.

[0068] Where e is the first error of the path tracking system, representing the lateral distance between the vehicle's center of gravity and the reference path; The second error of the path tracking system represents the vehicle's current heading angle. The deviation of the heading angle from the reference path; s is the distance the vehicle has traveled along the reference path; v x and v y These are the vehicle's longitudinal speed and lateral speed, respectively.

[0069] Considering the vehicle's dynamic characteristics and its nonlinear and uncertain dynamics, the vehicle's dynamic model is established as follows:

[0070] in, and The sideslip angle and yaw rate of the vehicle's center of gravity, respectively, F f and Fr These are the tire forces on the front and rear axles, respectively, where m is the mass of the vehicle, and I is the tire force on the front and rear axles, respectively. z Let l be the vehicle's moment of inertia about the Z-axis, and l1 and l2 be the distances from the vehicle's center of mass to the front and rear axles, respectively. These represent the unmodeled nonlinear and uncertain dynamics in vehicle lateral dynamics and yaw dynamics, respectively. and These represent the unmodeled nonlinear and uncertain dynamics in the front axle tire forces and rear axle tire forces, respectively.

[0071] Step 2: Construct the vehicle's comprehensive path tracking error The first error e and the second error This indicates that a comprehensive path tracking error model is established based on the vehicle's path tracking model and dynamics model, thereby transforming the complex path tracking problem into a convergence control problem of the comprehensive path tracking error.

[0072] The comprehensive path tracking error consists of a first error and a second error, and its form is as follows:

[0073] The overall path tracking error consists of a first error and a second error, x p The aiming distance is used as a guide. By integrating the path tracking error, the path tracking problem of autonomous vehicles is transformed into a convergence control problem of the integrated path tracking error.

[0074] Taking the derivative of the integrated path tracking error, we can obtain

[0075] Where, k r The curvature of the reference path, This is the approximation error resulting from the small-angle approximation of the first and second errors.

[0076] Further derivative of the integrated path tracking error yields...

[0077] Based on the vehicle's dynamics model, further results can be obtained

[0078] Guarantee the overall path tracking error e p Convergence enables autonomous vehicles to follow a reference path, thus transforming the complex path tracking problem into a convergence control problem of the comprehensive path tracking error.

[0079] Step 3: Through reasonable model transformation, a nonlinear extended state observer is constructed based on the observable output of the path tracking system and the transformed integrated path tracking error model to conduct real-time observation of the system's unmeasurable states, unmodeled nonlinearities, and uncertain disturbances.

[0080] The aforementioned reasonable model transformation of the integrated path tracking error system results in the integrated path tracking error system being expressed in the following form:

[0081] Viewing D(t) as the extended state x3 in the integrated path tracking error model, and utilizing the integrated path tracking error system output y, the measurable state variables of the vehicle... Including reference path information, a nonlinear extended state observer of the following form is established:

[0082] in, The observations of x1, x2, and x3 in the integrated path tracking error system are respectively Measured value. Control parameters and Each satisfies Control parameters It is a Hurwitz matrix, meaning there exists a positive definite matrix that satisfies The designed nonlinear extended state observer can monitor the synthesis path. The state variables in the tracking error system are observed in real time.

[0083] Step 4: Based on the path tracking system output, the nonlinear extended state observer state, and the disturbance observation values, design the integrated path tracking error system angle control law to compensate for the nonlinear dynamics in the path tracking error system and to cope with the uncertain disturbances in the system.

[0084] The rotation angle control rate of the integrated path tracking error system can be designed as follows:

[0085] This control law is designed based on the integrated path tracking error system output, the vehicle's measurable state variables, and the state estimates from the extended state observer. It enables the integrated path tracking error to converge to a minimum within a finite time. Applying this control law to the front wheel steering angle achieves high-precision path tracking control.

[0086] This control law is designed based on the output of the integrated path tracking error system, the vehicle's measurable state variables, and the state estimates from the extended state observer. It provides real-time compensation for nonlinear dynamics and uncertain disturbances in the integrated path tracking error system, enabling the vehicle's integrated path tracking error to converge to a minimum within a finite time. Applying this control law to the front wheel steering angle achieves high-precision path tracking control.

[0087] Step 5: Based on the nonlinear controller designed in the previous steps, path tracking control is performed. The vehicle's position and attitude information is input into the controller based on disturbance observation and compensation to obtain the steering angle control quantity.

[0088] By controlling the autonomous vehicle using the aforementioned steering angle control parameters, a better path tracking effect can be achieved.

[0089] The nonlinear path tracking model of the vehicle under planar motion in step one is shown below.

[0090]

[0091] Where e is the first error of the path tracking system, representing the lateral distance between the vehicle's center of gravity and the reference path; The second error of the path tracking system represents the vehicle's current heading angle. The deviation of the heading angle from the reference path; s is the distance the vehicle has traveled along the reference path; v x and v y These are the vehicle's longitudinal speed and lateral speed, respectively.

[0092] Considering the coupling nonlinearity, disturbances, and parameter uncertainties in the vehicle steering system, a dynamic model of the steering system is established.

[0093]

[0094] in, and The sideslip angle and yaw rate of the vehicle's center of gravity, respectively, F f and F r These are the tire forces on the front and rear axles, respectively, where m is the mass of the vehicle, and I is the tire force on the front and rear axles, respectively. z Let l be the vehicle's moment of inertia about the Z-axis, and l1 and l2 be the distances from the vehicle's center of mass to the front and rear axles, respectively. These represent the unmodeled nonlinear and uncertain dynamics in vehicle lateral dynamics and yaw dynamics, respectively. and These represent the unmodeled nonlinear and uncertain dynamics in the front axle tire forces and rear axle tire forces, respectively.

[0095] The comprehensive path tracking error of the vehicle constructed in step two is as follows:

[0096] The overall path tracking error consists of a first error and a second error, x p The aiming distance is used as a guide. By integrating the path tracking error, the path tracking problem of autonomous vehicles is transformed into a convergence control problem of the integrated path tracking error.

[0097] Taking the derivative of the integrated path tracking error, we can obtain

[0098] Where, k r Let be the curvature of the reference path, and let be the approximation error resulting from the small-angle approximation of the first and second errors.

[0099] Further derivative of the integrated path tracking error yields...

[0100] Based on the vehicle's dynamics model, further results can be obtained

[0101] The model conversion in step three represents the integrated path tracking error model in the following form.

[0102] Viewing D(t) as the extended state x3 in the integrated path tracking error model, and utilizing the integrated path tracking error system output y, the measurable state variables of the vehicle... Including reference path information, a nonlinear extended state observer of the following form is established:

[0103] in, The observations of x1, x2, and x3 in the integrated path tracking error system are respectively Measured value. Control parameters and Each satisfies Control parameters It is a Hurwitz matrix, meaning there exists a positive definite matrix that satisfies The designed nonlinear extended state observer can monitor the synthesis path. The state variables in the tracking error system are observed in real time.

[0104] The rotation angle control rate of the integrated path tracking error system described in step four can be designed as follows:

[0105] This control law is designed based on the integrated path tracking error system output, the vehicle's measurable state variables, and the state estimates from the extended state observer. It enables the integrated path tracking error to converge to a minimum within a finite time. Applying this control law to the front wheel steering angle achieves high-precision path tracking control.

[0106] Beneficial effects: (1) The control method described in this invention takes into account the strong coupling nonlinearity, uncertain environmental disturbances and time-varying characteristics of the autonomous vehicle path tracking system, and does not require the establishment of an accurate mathematical model, thus achieving high control accuracy and robustness.

[0107] (2) The control method described in this invention has a controller input that is only the state quantity that can be observed by the sensor. The algorithm is easy to implement in engineering and has strong practicality.

[0108] (3) The control method described in this invention is based on the vehicle path tracking kinematics and dynamics model for algorithm derivation, and has good applicability to both low-speed and high-speed vehicles.

[0109] (4) The control method described in this invention uses a nonlinear extended state observer to observe and compensate for nonlinear and uncertain disturbances in the system. Compared with the control method based on a linear extended state observer, it can better cope with nonlinearity and uncertainty in the system and achieve higher control accuracy and robustness.

[0110] To verify the effectiveness of the intelligent vehicle path tracking control method based on active disturbance rejection control proposed in this invention, a joint simulation test platform based on Matlab / Simulink and Carsim was established. Simulation results of pure tracking control, LQR control, and the active disturbance rejection control proposed in this invention at different vehicle speeds were compared under a dual-lane-change ideal reference path.

[0111] (1) The vehicle speed is 20km / h At a vehicle speed of 20 km / h, the path tracking results and lateral errors under the three control methods are as follows: Figure 4a and Figure 4b As shown. Figure 4b As shown, all three control methods exhibit good path tracking performance at low speeds. However, the results of the lateral error show that the lateral error of the active disturbance rejection control is the smallest, followed by the LQR control, and the pure tracking control is the largest. The maximum lateral errors of the three are 0.006m, 0.030m, and 0.033m, respectively.

[0112] (2) The vehicle speed is 40km / h At a vehicle speed of 40 km / h, the path tracking results and lateral errors under the three control methods are as follows: Figure 4c and Figure 4d As shown in the figure. The simulation results show that the active disturbance rejection control has the best path tracking performance, with a maximum lateral error of 0.038m; the LQR control still travels along the reference path, but the maximum lateral error is significantly improved compared to the low speed, reaching 0.091m; while the pure tracking control deviates far from the reference path at the curvature change point, with a maximum lateral error of 0.24m.

[0113] (3) The vehicle speed is 60km / h At a vehicle speed of 60 km / h, the path tracking results and lateral errors under the three control methods are as follows: Figure 4e and Figure 4f As shown in the figure, when the vehicle speed increases to 60 km / h, the nonlinear dynamic characteristics of the vehicle become more pronounced, and pure tracking control based on vehicle kinematics is no longer applicable, as it deviates significantly from the reference path. LQR control also deviates significantly from the reference path, especially at points where the curvature of the reference path changes. This is because the nonlinear dynamics of the vehicle are more pronounced during high-speed cornering, and the model error between the linear dynamics model and the actual vehicle dynamics is amplified. Controllers derived from linear models cannot cope with the effects of system nonlinearity and uncertainty. However, the active disturbance rejection control proposed in this invention still exhibits good path tracking performance, with a maximum lateral error of 0.13 m.

[0114] In summary, all three control methods exhibit good control performance at low speeds. The pure tracking control based on the kinematic model and the LQR control based on the linear dynamics model show a significant deterioration in control performance as vehicle speed increases. However, the LQR control, which takes into account vehicle dynamics, demonstrates better path tracking performance than the pure tracking control and has a wider applicable speed range. The active disturbance rejection controller design, which considers the nonlinearity and uncertainty of the system, shows good control performance at low, medium, and high speeds and exhibits strong robustness.

[0115] like Figure 5 The image shows a specific embodiment of an autonomous vehicle path tracking control device provided in this application. The device described in this embodiment is the physical device used to execute the method described in the above embodiments. Its technical solution is essentially the same as that of the above embodiments, and the corresponding descriptions in the above embodiments are also applicable to this embodiment. The device is applied to a business system, which includes a field-programmable gate array (FPGA) module and a storage module, wherein the storage module includes a first storage area and a second storage area; the device includes: Unit 501 is used to obtain the nonlinear path tracking model and dynamic model based on the vehicle's planar motion and dynamic characteristics. The second unit 502 is used to construct a dynamic model of the comprehensive path tracking error formed by the combination of lateral error and heading angle error and the aiming distance, based on the nonlinear path tracking model and the dynamic model; wherein, the dynamic model takes the comprehensive path tracking error as the system output and the front wheel steering angle control quantity as the system input. The third unit 503 is used to construct a nonlinear extended state observer based on the system output of the dynamic model, the measurable state variables of the vehicle, the reference path information of the vehicle, and the front wheel steering angle control quantity of the vehicle. The fourth unit 504 is used to obtain real-time estimates of unmeasurable states, nonlinear dynamics, and uncertain disturbances in the dynamic model using the nonlinear extended state observer. The fifth unit 505 is used to determine the front wheel steering angle control amount of the vehicle based on the real-time estimated value; The sixth unit 506 is used to control the vehicle according to the front wheel steering angle control amount.

[0116] Optionally, the process of obtaining the nonlinear path tracking model and dynamic model based on the vehicle's planar motion and dynamic characteristics includes: A two-degree-of-freedom vehicle dynamics model is established with the vehicle's center of gravity sideslip angle and yaw rate as state variables. This two-degree-of-freedom vehicle dynamics model is used as the dynamic model. The dynamic model is used to describe the dynamic relationship between the vehicle's lateral motion and yaw motion. The dynamic model includes unmodeled dynamic terms in the vehicle's lateral dynamics and yaw dynamics caused by model coupling nonlinearity, disturbances, and parameter uncertainties, as well as unmodeled dynamic terms in the front axle tire force and rear axle tire force. By using onboard sensors to acquire the vehicle's longitudinal velocity, lateral velocity, and heading angle in real time, and combining this with reference path information, a nonlinear path tracking model is constructed to describe the relationship between lateral error and heading angle error and the vehicle's motion.

[0117] Optionally, the step of constructing a dynamic model of the comprehensive path tracking error, which is formed by combining the lateral error and the heading angle error with the pre-aiming distance, based on the nonlinear path tracking model and the dynamic model, includes: The vehicle's current lateral error and current heading angle error are obtained, and the vehicle's current aiming distance is multiplied by the heading angle error and then added to the lateral error. The sum is used as the comprehensive path tracking error. The first and second derivatives of the time of the integrated path tracking error are obtained. In the process of differentiation, the derivatives of the lateral error and the heading angle error are replaced by the geometric kinematic relationship in the nonlinear path tracking model, and the small angle case is approximated. Substitute the dynamic relationship in the dynamic model, with the front wheel steering angle as input and the yaw rate and the center of gravity sideslip angle as state variables, into the terms related to the vehicle state change in the second derivative expression, so that the second derivative is finally expressed as the sum of the lumped unmodeled nonlinear and uncertain dynamic terms, the observable state terms, and the coefficient terms of the control gain multiplied by the front wheel steering angle control quantity. A dynamic model in the form of input-output is obtained, with the front wheel steering angle control quantity as input and the comprehensive path tracking error as output. This dynamic model is a second-order system and includes the lumped unmodeled nonlinear and uncertain dynamic terms. The lumped unmodeled nonlinear and uncertain dynamic terms include unmodeled dynamics caused by model coupling nonlinearities in vehicle lateral dynamics and yaw dynamics, external disturbances and parameter uncertainties, tire unmodeled dynamics, and the rate of change of small angle approximation error.

[0118] Optionally, the step of constructing a nonlinear extended state observer based on the system output of the dynamic model, the measurable state variables of the vehicle, the reference path information of the vehicle, and the front wheel steering angle control quantity of the vehicle includes: The dynamic model of the integrated path tracking error is transformed into a second-order system through variable substitution. The first state variable is defined as the integrated path tracking error, the second state variable is defined as the first derivative of the integrated path tracking error, and the lumped unmodeled nonlinear and uncertain dynamic terms in the dynamic model are expanded into a third state variable. The third state variable is used as the total disturbance to form a third-order extended state system. The measured value of the integrated path tracking error is used as the system output. A nonlinear extended state observer is constructed based on the third-order extended state system. The observer takes the system output and the front wheel steering angle control quantity as inputs and introduces nonlinear functions and gain parameters to update the estimated values ​​of each state variable online.

[0119] Optionally, the observer updates its estimates of each state variable according to the following rules: The rate of change of the estimated value of the first state variable is equal to the sum of the estimated value of the second state variable and the first nonlinear term. The first nonlinear term is obtained by dividing the first gain parameter by the cube of the preset scaling parameter and then multiplying it by the value of the observation error after the first nonlinear transformation. The observation error is the difference between the system output and the estimated value of the first state variable. The rate of change of the estimated value of the second state variable is equal to the sum of the first and second time-varying terms, the estimated value of the third state variable, the product of the coefficient term and the front wheel steering angle control quantity, and the second nonlinear term, which can be calculated using the measurable state variables of the vehicle and the reference path information. The second nonlinear term is obtained by dividing the second gain parameter by the preset scaling parameter and then multiplying the observation error by the value after the second nonlinear transformation. The rate of change of the estimated value of the third state variable is equal to the third gain parameter multiplied by the observation error after the third nonlinear transformation; The first to third nonlinear transformations all follow this rule: the output after the nonlinear transformation of the input value is equal to the absolute value of the input value raised to a specified power multiplied by the sign function value of the input value. The specified powers used in the first to third nonlinear transformations are all different, and each specified power is calculated according to a common preset base parameter and corresponding rules. The values ​​of the first to third gain parameters ensure that the observer error dynamic matrix is ​​a Herwitz matrix, where the Herwitz matrix is... , , , These are the first gain parameter, the second gain parameter, and the third gain parameter, respectively.

[0120] Optionally, obtaining real-time estimates of unmeasurable states, nonlinear dynamics, and uncertain disturbances in the dynamic model using the nonlinear extended state observer includes: In each control cycle, the first state variable estimate of the output of the nonlinear extended state observer is obtained as the real-time estimate of the integrated path tracking error. The second state variable estimate output by the nonlinear extended state observer is used as the real-time estimate of the unmeasurable state in the dynamic model, where the unmeasurable state is the first derivative of the real-time estimate of the integrated path tracking error. The estimated value of the third state variable, which is estimated and output in real time by the nonlinear extended state observer based on its preset update rule, is used as the real-time estimate of the unmodeled nonlinear dynamics and uncertain disturbances in the dynamic model. The unmodeled nonlinear dynamics and uncertain disturbances ensemble includes unmodeled dynamics, parameter uncertainties, and external disturbances.

[0121] Optionally, determining the front wheel steering angle control value of the vehicle based on the real-time estimated value includes: Using the vehicle's longitudinal acceleration, lateral acceleration, yaw rate, heading angle error, longitudinal velocity, and reference path curvature information, the first time-varying feedforward term and the second time-varying feedforward term are calculated in real time. The front wheel steering angle control quantity is calculated based on the real-time estimate of the total disturbance output by the nonlinear extended state observer, the first time-varying feedforward term and the second time-varying feedforward term, and the preset feedback term related to the scaling parameter and the observer gain, in a manner that combines feedforward compensation, disturbance compensation and error feedback. The feedback term is composed of the measured value of the integrated path tracking error multiplied by a first feedback coefficient, plus the real-time estimated value of the rate of change of the integrated path tracking error multiplied by a second feedback coefficient. The first feedback coefficient and the second feedback coefficient are determined by the square of the scaling parameter and the product of the scaling parameter and the preset stability parameter, respectively.

[0122] Optionally, the expression for calculating the front wheel steering angle control quantity is as follows: take the negative of the sum of the first time-varying feedforward term and the second time-varying feedforward term, subtract the real-time estimate of the total disturbance, add the feedback term, and divide the result by the control gain coefficient related to the vehicle parameters and the aiming distance to obtain the front wheel steering angle control quantity; wherein, the control gain coefficient is determined by the front axle tire lateral stiffness, the distance from the vehicle center of gravity to the front axle, the vehicle's moment of inertia about the vertical axis, and the aiming distance.

[0123] Optionally, before the step of controlling the vehicle according to the front wheel steering angle control amount, the method further includes: The calculated front wheel steering angle control value is subjected to amplitude limiting processing. If the front wheel steering angle control value exceeds the preset maximum steering angle amplitude, the front wheel steering angle control value is set to the maximum steering angle amplitude. The rate of change of the front wheel steering angle control value after amplitude limiting is applied; If the time change rate of the front wheel steering angle control quantity after amplitude limiting exceeds the preset maximum change rate, then the change rate of the front wheel steering angle control quantity after amplitude limiting will be limited to the maximum change rate. The preset maximum steering angle amplitude and the preset maximum rate of change are preset based on the physical constraints of the vehicle steering system.

[0124] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. At the hardware level, the electronic device includes a processor, and optionally also includes an internal bus, a network interface, and a memory. The memory may include RAM, such as high-speed random-access memory (RAM), or non-volatile memory, such as at least one disk storage device. Of course, the electronic device may also include other hardware required for other services.

[0125] Memory is used to store instructions for execution. Specifically, instructions for execution are computer programs that can be executed. Memory can include main memory and non-volatile memory, and it provides the processor with execution instructions and data.

[0126] In one possible implementation, the processor reads the corresponding execution instructions from non-volatile memory into main memory and then executes them. Alternatively, it may obtain the corresponding execution instructions from other devices to form an autonomous vehicle path tracking control device at the logical level. The processor executes the execution instructions stored in the memory to implement the autonomous vehicle path tracking control method provided in any embodiment of this application.

[0127] The above is as stated in this application. Figure 1 The method executed by the autonomous vehicle path tracking control device provided in the illustrated embodiment can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor.

[0128] The steps of the method disclosed in the embodiments of this application can be directly manifested as being executed by a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0129] This application also proposes a readable medium that stores execution instructions. When the stored execution instructions are executed by the processor of an electronic device, the electronic device can execute the autonomous vehicle path tracking control method provided in any embodiment of this application, and specifically perform the above-mentioned evaluation method.

[0130] The electronic devices described in the foregoing embodiments may be computers.

[0131] Those skilled in the art will understand that the embodiments of this application can be provided as methods or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or a combination of software and hardware.

[0132] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0133] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0134] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A path tracking control method for an autonomous vehicle, characterized in that, Includes the following steps: Based on the vehicle's planar motion and dynamic characteristics, a nonlinear path tracking model and a dynamic model are obtained; Based on the nonlinear path tracking model and the dynamic model, a dynamic model of the comprehensive path tracking error, which is formed by the combination of lateral error and heading angle error with the aiming distance, is constructed; wherein, the dynamic model takes the comprehensive path tracking error as the system output and the front wheel steering angle control quantity as the system input; Based on the system output of the dynamic model, the measurable state variables of the vehicle, the reference path information of the vehicle, and the front wheel steering angle control quantity of the vehicle, a nonlinear extended state observer is constructed. The nonlinear extended state observer is used to obtain real-time estimates of unmeasurable states, nonlinear dynamics, and uncertain disturbances in the dynamic model; Based on the real-time estimated value, the front wheel steering angle control amount of the vehicle is determined; The vehicle is controlled based on the front wheel steering angle control value.

2. The method according to claim 1, characterized in that, Based on the vehicle's planar motion and dynamic characteristics, a nonlinear path tracking model and a dynamic model are obtained, including: A two-degree-of-freedom vehicle dynamics model is established with the vehicle's center of gravity sideslip angle and yaw rate as state variables. This two-degree-of-freedom vehicle dynamics model is used as the dynamic model. The dynamic model is used to describe the dynamic relationship between the vehicle's lateral motion and yaw motion. The dynamic model includes unmodeled dynamic terms in the vehicle's lateral dynamics and yaw dynamics caused by model coupling nonlinearity, disturbances, and parameter uncertainties, as well as unmodeled dynamic terms in the front axle tire force and rear axle tire force. By using onboard sensors to acquire the vehicle's longitudinal velocity, lateral velocity, and heading angle in real time, and combining this with reference path information, a nonlinear path tracking model is constructed to describe the relationship between lateral error and heading angle error and the vehicle's motion.

3. The method according to claim 1, characterized in that, The construction of a dynamic model for the comprehensive path tracking error, based on the nonlinear path tracking model and the dynamic model, and combining the lateral error, heading angle error, and pre-aiming distance, includes: The vehicle's current lateral error and current heading angle error are obtained, and the vehicle's current aiming distance is multiplied by the heading angle error and then added to the lateral error. The sum is used as the comprehensive path tracking error. The first and second derivatives of the time of the integrated path tracking error are obtained. In the process of differentiation, the derivatives of the lateral error and the heading angle error are replaced by the geometric kinematic relationship in the nonlinear path tracking model, and the small angle case is approximated. Substitute the dynamic relationship in the dynamic model, with the front wheel steering angle as input and the yaw rate and the center of gravity sideslip angle as state variables, into the terms related to the vehicle state change in the second derivative expression, so that the second derivative is finally expressed as the sum of the lumped unmodeled nonlinear and uncertain dynamic terms, the observable state terms, and the coefficient terms of the control gain multiplied by the front wheel steering angle control quantity. A dynamic model in input-output form is obtained, with the front wheel steering angle control quantity as input and the comprehensive path tracking error as output. This dynamic model is a second-order system and includes the lumped unmodeled nonlinear and uncertain dynamic terms. The lumped unmodeled nonlinear and uncertain dynamic terms include unmodeled dynamics caused by model coupling nonlinearities in vehicle lateral dynamics and yaw dynamics, external disturbances and parameter uncertainties, tire unmodeled dynamics, and the rate of change of the approximation error.

4. The method according to claim 3, characterized in that, The construction of a nonlinear extended state observer based on the system output of the dynamic model, the measurable state variables of the vehicle, the reference path information of the vehicle, and the front wheel steering angle control value of the vehicle includes: The dynamic model of the integrated path tracking error is transformed into a second-order system through variable substitution. The first state variable is defined as the integrated path tracking error, the second state variable is defined as the first derivative of the integrated path tracking error, and the lumped unmodeled nonlinear and uncertain dynamic terms in the dynamic model are expanded into a third state variable. The third state variable is used as the total disturbance to form a third-order extended state system. The measured value of the integrated path tracking error is used as the system output. A nonlinear extended state observer is constructed based on the third-order extended state system. The observer takes the system output and the front wheel steering angle control quantity as inputs and introduces nonlinear functions and gain parameters to update the estimated values ​​of each state variable online.

5. The method according to claim 4, characterized in that, The observer updates its estimates of each state variable according to the following rules: The rate of change of the estimated value of the first state variable is equal to the sum of the estimated value of the second state variable and the first nonlinear term. The first nonlinear term is obtained by dividing the first gain parameter by the cube of the preset scaling parameter and then multiplying it by the value of the observation error after the first nonlinear transformation. The observation error is the difference between the system output and the estimated value of the first state variable. The rate of change of the estimated value of the second state variable is equal to the sum of the first and second time-varying terms, the estimated value of the third state variable, the product of the coefficient term and the front wheel steering angle control quantity, and the second nonlinear term, which can be calculated using the measurable state variables of the vehicle and the reference path information. The second nonlinear term is obtained by dividing the second gain parameter by the preset scaling parameter and then multiplying the observation error by the value after the second nonlinear transformation. The rate of change of the estimated value of the third state variable is equal to the third gain parameter multiplied by the observation error after the third nonlinear transformation; The first to third nonlinear transformations all follow this rule: the output after the nonlinear transformation of the input value is equal to the absolute value of the input value raised to a specified power multiplied by the sign function value of the input value. The specified powers used in the first to third nonlinear transformations are all different, and each specified power is calculated according to a common preset base parameter and corresponding rules. The values ​​of the first to third gain parameters ensure that the observer error dynamic matrix is ​​a Herwitz matrix, where the Herwitz matrix is... , , , These are the first gain parameter, the second gain parameter, and the third gain parameter, respectively.

6. The method according to claim 1, characterized in that, The step of obtaining real-time estimates of unmeasurable states, nonlinear dynamics, and uncertain disturbances in the dynamic model using the nonlinear extended state observer includes: In each control cycle, the first state variable estimate of the output of the nonlinear extended state observer is obtained as the real-time estimate of the integrated path tracking error. The second state variable estimate output by the nonlinear extended state observer is used as the real-time estimate of the unmeasurable state in the dynamic model, where the unmeasurable state is the first derivative of the real-time estimate of the integrated path tracking error. The estimated value of the third state variable, which is estimated and output in real time by the nonlinear extended state observer based on its preset update rule, is used as the real-time estimate of the unmodeled nonlinear dynamics and uncertain disturbances in the dynamic model. The unmodeled nonlinear dynamics and uncertain disturbances ensemble includes unmodeled dynamics, parameter uncertainties, and external disturbances.

7. The method according to claim 5, characterized in that, Determining the front wheel steering angle control value of the vehicle based on the real-time estimated value includes: Using the vehicle's longitudinal acceleration, lateral acceleration, yaw rate, heading angle error, longitudinal velocity, and reference path curvature information, the first and second time-varying feedforward terms are calculated in real time. The front wheel steering angle control quantity is calculated based on the real-time estimate of the total disturbance output by the nonlinear extended state observer, the first time-varying feedforward term and the second time-varying feedforward term, and the preset feedback term related to the scaling parameter and the observer gain, in a manner that combines feedforward compensation, disturbance compensation and error feedback. The feedback term is composed of the measured value of the integrated path tracking error multiplied by a first feedback coefficient, plus the real-time estimated value of the rate of change of the integrated path tracking error multiplied by a second feedback coefficient. The first feedback coefficient and the second feedback coefficient are determined by the square of the scaling parameter and the product of the scaling parameter and the preset stability parameter, respectively.

8. The method according to claim 7, characterized in that, The expression for calculating the front wheel steering angle control quantity is as follows: take the negative of the sum of the first time-varying feedforward term and the second time-varying feedforward term, subtract the real-time estimate of the total disturbance, add the feedback term, and divide the result by the control gain coefficient related to the vehicle parameters and the aiming distance to obtain the front wheel steering angle control quantity; wherein, the control gain coefficient is determined by the front axle tire lateral stiffness, the distance from the vehicle's center of gravity to the front axle, the vehicle's moment of inertia about the vertical axis, and the aiming distance.

9. The method according to claim 1, characterized in that, Prior to the step of controlling the vehicle based on the front wheel steering angle control amount, the method further includes: The calculated front wheel steering angle control value is subjected to amplitude limiting processing. If the front wheel steering angle control value exceeds the preset maximum steering angle amplitude, the front wheel steering angle control value is set to the maximum steering angle amplitude. The rate of change of the front wheel steering angle control value after amplitude limiting is applied; If the time change rate of the front wheel steering angle control quantity after amplitude limiting exceeds the preset maximum change rate, then the change rate of the front wheel steering angle control quantity after amplitude limiting will be limited to the maximum change rate. The preset maximum steering angle amplitude and the preset maximum rate of change are preset based on the physical constraints of the vehicle steering system.

10. A path tracking control device for an autonomous vehicle, characterized in that, The device includes: The first unit is used to obtain the nonlinear path tracking model and the dynamic model based on the vehicle's planar motion and dynamic characteristics; The second unit is used to construct a dynamic model of the comprehensive path tracking error formed by the combination of lateral error and heading angle error and the aiming distance, based on the nonlinear path tracking model and the dynamic model; wherein, the dynamic model takes the comprehensive path tracking error as the system output and the front wheel steering angle control quantity as the system input. The third unit is used to construct a nonlinear extended state observer based on the system output of the dynamic model, the measurable state variables of the vehicle, the reference path information of the vehicle, and the front wheel steering angle control quantity of the vehicle. The fourth unit is used to obtain real-time estimates of unmeasurable states, nonlinear dynamics, and uncertain disturbances in the dynamic model using the nonlinear extended state observer. The fifth unit is used to determine the front wheel steering angle control amount of the vehicle based on the real-time estimated value; The sixth unit is used to control the vehicle based on the front wheel steering angle control amount.

Images