Self-adaptive vehicle tracking control method based on main ring-servo ring layered architecture

Abstract

The invention discloses a self-adaptive vehicle tracking control method based on a main ring-servo ring layered architecture, which comprises a main ring control level and a servo ring control level, and is characterized in that the main ring control level is used for carrying out online planning on an ideal driving track of vehicle driving to obtain the ideal driving track; and the servo ring level is used for calculating an ideal control input signal of a corresponding actuator according to the ideal driving track output by the main ring control level and the interface form of the EPS actuator of the vehicle so as to realize tracking control of the vehicle on the ideal driving track. The tracking control method has the advantages that the state of the vehicle can be detected in real time, and the vehicle is smoothly controlled to track an ideal driving track in a safety domain; the algorithm adopts a layered modular design, each core function module can flexibly select different feasible technical schemes to be combined according to customer project requirements and system software and hardware conditions, and the adaptability is high.

Description

Technical Field

[0001] The present invention relates to the field of assisted driving, and in particular to the field of forward control based on vehicle target tracking trajectory in assisted driving. Background Art

[0002] ADAS systems are designed to enhance the driving experience and assist in vehicle control. One such function is trajectory tracking technology, which is also known as vehicle tracking control. Based on lane recognition by cameras, the system adjusts the vehicle's position relative to the lane lines through the EPS actuator interface (steering wheel angle / torque or front wheel angle) to ensure the vehicle tracks the target trajectory.

[0003] Existing vehicle trajectory tracking technology suffers from shortcomings such as low system tracking accuracy, poor robustness, and difficulty adapting to different vehicle models and their actuators. The former affects the safety and comfort of vehicle centering control, reducing drivers' trust in and satisfaction with the ADAS system. It can even lead to poor vehicle control in some extreme conditions (such as negotiating high-speed curves), creating the risk of collisions with guardrails or other traffic participants in adjacent lanes. The latter, on the other hand, weakens the algorithm's universality and increases the time and labor costs of R&D and project adaptation.

[0004] Therefore, designing a vehicle tracking algorithm with a clear architecture, precise control, strong robustness, and the ability to flexibly adapt to different actuator requirements can improve the overall control performance of the ADAS controller, thereby enhancing the user's driving experience and facilitating adaptation to different project requirements. Summary of the Invention

[0005] The purpose of the present invention is to overcome the shortcomings of the existing technology and provide an adaptive vehicle tracking control method based on a main loop-servo loop layered architecture to achieve smoother response and higher tracking accuracy, and can be adapted to different EPS actuators.

[0006] In order to achieve the above objectives, the technical solution adopted by the present invention is: a vehicle tracking control method based on a main loop-servo loop layered architecture, including a main loop control layer and a servo loop control layer,

[0007] The main loop control layer is used to plan the ideal driving trajectory of the vehicle online to obtain the ideal driving trajectory;

[0008] The servo loop layer is used to calculate the ideal control input signal of the corresponding actuator based on the ideal driving trajectory output by the main loop control layer and the vehicle EPS actuator interface form, so as to achieve vehicle tracking control of the ideal driving trajectory; an adaptive algorithm is used to compensate the measurement values ​​of the vehicle steering wheel angle sensor and vehicle lateral acceleration sensor based on the vehicle state to improve the robustness of the algorithm.

[0009] The main loop control layer includes an ideal trajectory planning module and a vehicle dynamics safety boundary estimation module. Based on the lane line scene obtained by visual recognition, the ideal driving trajectory of the vehicle is planned online according to the lateral control function requirements based on the real-time estimation of the vehicle dynamics boundary.

[0010] The ideal trajectory planning module plans the ideal driving trajectory based on lane line information and vehicle information; the vehicle dynamics safety margin estimation module obtains the corresponding vehicle dynamics safety boundary based on the ideal driving trajectory and vehicle model, and adjusts the ideal vehicle trajectory information output by the ideal trajectory planning module based on the safety boundary information.

[0011] The ideal trajectory planning module uses the following method to plan the trajectory:

[0012] The four-dimensional vehicle initial state X0 = [x0, y0, θ0, κ0] is constructed based on the vehicle's real-time longitudinal and lateral positions, heading angles, and curve curvature in the lane. Similarly, the vehicle's planned terminal state X0 can be defined by selecting an appropriate preview time / distance based on the vehicle speed. F =[x0+x pre ,y f ,θ f , κ f For a set [X0, X F ], the geometric path is generated based on the following kinematic model: x(s) = ∫cos(θ(s))ds, y(s) = ∫sin(θ(s))ds, θ(s) = ∫κ(s)ds, κ(s) = ∑κ i s i , where s is the distance along the center line of the lane;

[0013] For lateral control, the vehicle speed plan V(s) can be set as a constant speed based on the vehicle safety boundary as a constraint condition.

[0014] The vehicle dynamics safety margin estimation module is implemented using the following method:

[0015] Recommended vehicle dynamics model,Based on the above vehicle dynamics model, the following vehicle stability,bounds can be defined:|a y |≤a y,max ,in is the lateral acceleration of the vehicle, a y,max is the maximum lateral acceleration threshold defined based on comfort and safety, V and r are the lateral acceleration, vehicle speed, and yaw rate of the vehicle body measured by the sensors, respectively. It is recommended that r satisfy the following stability constraints:

[0016] -(l F +lR )g / [V 2 -(wr / 2 )2]<K US <2(l F +l R )g / [V 2 -(wr / 2) 2 ],

[0017] where K US =mg(l R C αR -l F C αF ) / [(l F +l R )C αF C αR ] is the newly defined vehicle understeer coefficient, l F and l R are the distances from the vehicle's center of mass to the front and rear axles, C αF and C αR are the front axle cornering stiffness and rear axle cornering stiffness of the vehicle respectively, m is the vehicle mass, g is the acceleration of gravity, and w is the left and right wheelbase of the vehicle.

[0018] The real-time position and posture of the vehicle are obtained through the camera, and the preview point is selected as the control target in the ideal driving trajectory output by the upper main loop. The control algorithm based on geometric relationships or dynamics, or the model-free control algorithm is used to track the ideal driving trajectory.

[0019] The advantages of the present invention are: the tracking control method can detect the vehicle status in real time and smoothly control the vehicle to track the ideal driving trajectory within the safety domain; the algorithm adaptively compensates the measurement values ​​of the vehicle steering wheel angle sensor and the vehicle lateral acceleration sensor, which can improve the robustness of tracking control; the algorithm adopts a layered modular design, and each core functional module can flexibly select different feasible technical solutions for combination according to customer project requirements and system hardware and software conditions, and has strong adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The following is a brief description of the contents and symbols in the drawings of the present invention:

[0021] Figure 1 Schematic diagram of the control method of the present invention;

[0022] Figure 2 Schematic diagram of a tire model for the control method of the present invention. DETAILED DESCRIPTION

[0023] The specific implementation of the present invention will be further explained in detail below by describing the best embodiment with reference to the accompanying drawings.

[0024] The main purpose of this invention is to address the control accuracy problems and the problem of flexible adaptation to different actuator requirements of the tracking control algorithm in the existing technology, and to design a vehicle tracking algorithm with clear architecture, precise control, and flexible adaptation to different actuator requirements. It can improve the overall control performance of the ADAS controller, thereby enhancing the user's driving experience, and is conducive to adapting to different project requirements.

[0025] like Figure 1 As shown, a vehicle tracking control method based on a main loop-servo loop layered architecture includes a main loop control layer and a servo loop control layer.

[0026] The main loop control layer is used to plan the ideal driving trajectory of the vehicle online to obtain the ideal driving trajectory;

[0027] The servo loop layer calculates the ideal control input signals for the corresponding actuators based on the ideal trajectory output by the main control layer and the vehicle's EPS actuator interface format, thereby enabling the vehicle to track the ideal trajectory. The main control layer includes an ideal trajectory planning module and a vehicle dynamics safety margin estimation module. Based on the lane line scene obtained by visual recognition and real-time estimation of the vehicle dynamics margin, the ideal trajectory is planned online according to the requirements of the lateral control function.

[0028] The ideal trajectory planning module plans the ideal driving trajectory based on lane line information and vehicle information; the vehicle dynamics safety margin estimation module obtains the corresponding vehicle dynamics safety boundary based on the ideal driving trajectory and vehicle model, and adjusts the ideal vehicle trajectory information output by the ideal trajectory planning module based on the safety boundary information.

[0029] Specifically, a layered vehicle trajectory tracking algorithm is designed, with the algorithm architecture divided into two levels: main-loop control and servo-loop control. The upper-level main-loop algorithm consists of an ideal trajectory planning module and a vehicle dynamics safety boundary estimation module. Based on the lane line scene obtained by visual recognition, the ideal driving trajectory of the vehicle is planned online based on the real-time estimation of the vehicle dynamics boundary and the requirements of the lateral control function (driving straight or along a curve, staying in the center or leaning to one side, etc.). Based on the modular design concept, different trajectory planning methods and vehicle and tire models with different degrees of freedom can be combined according to conditions such as hardware computing power and the characteristics of the controlled object.

[0030] The lower-level servo loop control algorithm calculates the ideal control input of the actuator based on the ideal driving trajectory obtained by the upper-level main loop control algorithm (depending on the actuator interface, it can be selected between the steering wheel angle / torque or the vehicle's front wheel angle), thereby achieving vehicle tracking control of the ideal driving trajectory. At the same time, an adaptive algorithm is used to compensate the measurement values ​​of the vehicle steering wheel angle sensor and vehicle lateral acceleration sensor based on the vehicle state to improve the robustness of the algorithm.

[0031] The logic block diagram of the algorithm designed by the present invention is shown in the attached Figure 1 , the specific implementation method is as follows:

[0032] 1) Ideal trajectory planning: The four-dimensional vehicle initial state X0 = [x0, y0, θ0, κ0] is constructed based on the vehicle's real-time longitudinal and lateral position in the lane, its heading angle, and the curvature of the curve. (x0, y0) are the vehicle's coordinates in the vehicle coordinate system (where x points in the direction of travel and y points to the left), θ0 is the heading angle of the lane in the vehicle coordinate system, and κ0 is the curvature of the lane in the vehicle coordinate system. Similarly, by selecting an appropriate preview time / distance based on vehicle speed, the planned vehicle endpoint state X0 can be defined. F =[x0+x pre ,y f ,θ f , κ f ], where x pre 、y f are the longitudinal distance and lateral distance of the preview point, θ f , κ f They represent the orientation angle and curvature of the lane where the preview point is located (i.e., the lane adjacent to the lane change). F ], the geometric path is generated based on the following kinematic model: x(s) = ∫cos(θ(s))ds, y(s) = ∫sin(θ(s))ds, θ(s) = ∫κ(s)ds, κ(s) = ∑κ i s i , where x and y are the longitudinal and lateral distances of a point on the planned path in the vehicle coordinate system, s is the distance along the centerline of the lane, θ and κ represent the heading angle and curvature of the planned path at that point, respectively, and κ i is the polynomial coefficient, i is the order of the curvature polynomial model. F The geometric path is given by the polynomial coefficient κ i Uniquely determined. i It can be solved by the following 2-PBV (Two-point boundary value) iterative algorithm: P(k+1)=P(k)+ΔP. For ADAS lateral control, the vehicle speed planning V(s) can be set as a constant speed based on the vehicle safety boundary as a constraint condition;

[0033] 2) Vehicle dynamics boundary: According to the different characteristics of the controlled object (rollover tendency), a two-degree-of-freedom (2-DOF) or three-degree-of-freedom (3-DOF) vehicle model can be selected, where the 2-DOF vehicle model is Among them, l F 、l R are the distances from the vehicle's center of mass to the front axle and rear axle, m F 、m R are the loads on the front and rear axles respectively, r is the yaw rate of the vehicle, m is the mass of the vehicle, a y is the lateral acceleration of the vehicle, I zz is the yaw moment of inertia of the vehicle;

[0034] The 3-DOF vehicle model is Among them, F y is the lateral force of the vehicle along the y-axis, M z 、M x They represent yaw torque and roll torque respectively, h b is the height from the center of mass of the sprung mass to the roll center, m b is the sprung mass, φ is the roll angle of the vehicle, I xx is the vehicle's roll moment of inertia, D φ is the roll damping of the vehicle, K φ is the roll stiffness of the vehicle, and g is the acceleration due to gravity.

[0035] like Figure 2 As shown, in order to reduce the complexity of the algorithm, a linear tire model F can be used. y =C α α, C α is the tire cornering stiffness, α=(v+l F r) / V-δ F is the front axle slip angle, where v is the lateral speed of the vehicle along the y-axis, V represents the vehicle speed, and δ F is the front wheel turning angle of the vehicle. Based on the above vehicle dynamics model, the following vehicle stability boundary can be defined: |a y |≤a y，max ,in is the lateral acceleration of the vehicle, a y，max is the maximum lateral acceleration threshold defined based on comfort and safety, V and r are the vehicle lateral acceleration, vehicle speed, and vehicle yaw rate measured by the sensors, respectively. It is recommended that r satisfy the following stability constraints: -(l F +lR )g / [V 2 -(wr / 2) 2 ]<K US <2(l F +l R )g / [V 2 -(wr / 2) 2 ],

[0036] where K US =mg(l R C αR -l F C αF ) / [(l F +l R )C αF C αR ] is the newly defined vehicle understeer coefficient, l F and l R are the distances from the vehicle's center of mass to the front and rear axles, C αF and C αR are the front axle cornering stiffness and rear axle cornering stiffness of the vehicle respectively, m is the vehicle mass, g is the acceleration of gravity, and w is the left and right wheelbase of the vehicle.

[0037] The vehicle dynamics boundary is fed back into the ideal trajectory planning module as a constraint condition for the planning problem to ensure the dynamic safety of the planned path: Assuming a constant vehicle speed, the above vehicle dynamics boundary can be obtained by a y =Vr converted to a y -r phase plane handling stability boundary; using the phase plane analysis method, when the vehicle state approaches the boundary, it is determined that the vehicle has an instability risk, and then the alternative lane change trajectory with instability risk is deleted from the candidate pool;

[0038] 3) EPS actuator servo control: The real-time position and posture of the vehicle is obtained through the camera, and the appropriate preview point is selected as the control target in the ideal driving trajectory output by the upper main loop. The control algorithm based on geometric relationship (for example: Ackerman angle, PurePursuit), kinematics (for example: kinematic bicycle model) or dynamics (for example: MPC), or model-free control algorithm (for example: PID) can be used to achieve the tracking of the ideal driving trajectory. Based on the above-mentioned hierarchical tracking algorithm design, the lateral position of the vehicle can be safely and smoothly corrected in real time to achieve the tracking of the ideal driving trajectory. The road curvature information is obtained through the camera, and the vehicle steering wheel angle sensor measurement error △δ is combined with the vehicle kinematic state and the driver (and / or ADAS controller) steering torque input. SWA and the vehicle lateral acceleration sensor measurement error △a y Perform real-time estimation and then use the following feedback integral control algorithm to adaptively compensate for the error:

[0039]

[0040] in, and are the steering wheel angle and vehicle lateral acceleration measured by the sensor, W δSWA and W ay They are the weight coefficients that control the speed of self-learning;

[0041] Obviously, the specific implementation of the present invention is not limited to the above-mentioned methods. As long as various non-substantial improvements are made using the method concept and technical solution of the present invention, they are all within the scope of protection of the present invention.

Claims

1. An adaptive vehicle tracking control method based on a main loop-servo loop layered architecture, characterized by: Including the main loop control level and the servo loop control level, The main loop control layer is used to plan the ideal driving trajectory of the vehicle online to obtain the ideal driving trajectory; The servo loop layer is used to calculate the ideal control input signal of the corresponding actuator based on the ideal driving trajectory output by the main loop control layer, the vehicle steering wheel angle and vehicle lateral acceleration information compensated by the adaptive algorithm, and the vehicle EPS actuator interface form to achieve vehicle tracking control of the ideal driving trajectory.

2. The adaptive vehicle tracking control method based on a main loop-servo loop layered architecture according to claim 1, wherein: The main loop control layer includes an ideal trajectory planning module and a vehicle dynamics safety boundary estimation module. Based on the lane line scene obtained by visual recognition, the ideal driving trajectory of the vehicle is planned online according to the lateral control function requirements based on the real-time estimation of the vehicle dynamics boundary.

3. The adaptive vehicle tracking control method based on a main loop-servo loop layered architecture according to claim 2, wherein: The ideal trajectory planning module plans the ideal driving trajectory based on lane information and vehicle information. The vehicle dynamics safety margin estimation module obtains the corresponding vehicle dynamics safety margin based on the ideal driving trajectory and vehicle model, and adjusts the ideal vehicle trajectory information output by the ideal trajectory planning module based on the safety margin information. An adaptive algorithm is used to compensate the measurements of the vehicle steering wheel angle sensor and the vehicle lateral acceleration sensor based on the vehicle state to improve the robustness of the algorithm.

4. The adaptive vehicle tracking control method based on a main loop-servo loop layered architecture according to any one of claims 1 to 3, characterized in that: The ideal trajectory planning module uses the following method to plan the trajectory: The four-dimensional vehicle initial state X0 = [x0, y0, θ0, κ0] is constructed based on the vehicle's real-time longitudinal and lateral positions, heading angles, and curve curvature in the lane. Similarly, the vehicle's planned terminal state X0 can be defined by selecting an appropriate preview time / distance based on the vehicle speed. F =[x0+x pre ,y f ,θ f , κ f For a set [X0, X F ], the geometric path is generated based on the following kinematic model: x(s) = ∫fos(θ(s))ds, y(s) = ∫sin(θ(s))ds, θ(s) = ∫κ(s)ds, κ(s) = ∑κ i s i , where s is the distance along the center line of the lane; For lateral control, the vehicle speed plan V(s) can be set as a constant speed based on the vehicle safety boundary as a constraint condition.

5. The adaptive vehicle tracking control method based on a main loop-servo loop layered architecture according to any one of claims 1 to 3, characterized in that: The vehicle dynamics safety margin estimation module is implemented using the following method: Establish a vehicle dynamics model. Based on the above vehicle dynamics model, the following vehicle stability boundary can be defined: y |≤a y,max ,in is the lateral acceleration of the vehicle, a y,max is the maximum lateral acceleration threshold defined based on comfort and safety, V and r are the lateral acceleration, vehicle speed, and yaw rate of the vehicle body measured by the sensors, respectively. It is recommended that r satisfy the following stability constraints: -(l F +l R )g / [V 2 -(wr / 2) 2 ]<K US <2(l F +l R )g / (V 2 -(wr / 2) 2 ], where K US =mg(l R C αR -l F C αF ) / [(l F +l R )C αF C αR ] is the newly defined vehicle understeer coefficient, l F and l R are the distances from the vehicle's center of mass to the front and rear axles, C αF and C αR are the front axle lateral stiffness and rear axle lateral stiffness of the vehicle respectively, m is the vehicle mass, g is the acceleration of gravity, and w is the left and right wheelbase of the vehicle.

6. The adaptive vehicle tracking control method based on a main loop-servo loop layered architecture according to any one of claims 1 to 3, characterized in that: The real-time position and posture of the vehicle are obtained through the camera, and the preview point is selected as the control target in the ideal driving trajectory output by the upper main loop. The control algorithm based on geometric relationships or dynamics, or the model-free control algorithm is used to track the ideal driving trajectory.

7. The adaptive vehicle tracking control method based on a main loop-servo loop layered architecture according to any one of claims 1 to 3, characterized in that: The road curvature information is obtained through the camera, and the vehicle steering wheel angle sensor measurement error △δ is calculated based on the vehicle kinematic state and the steering torque input from the driver and / or ADAS controller. SWA and the vehicle lateral acceleration sensor measurement error △a y Perform real-time estimation and then use the following feedback integral control algorithm to adaptively compensate for the error: in, and are the steering wheel angle and vehicle lateral acceleration measured by the sensor, W δSWA and W ay are the weight coefficients that control the speed of self-learning.

8. The adaptive vehicle tracking control method based on a main loop-servo loop layered architecture according to claim 5, wherein: A linear tire model is used as the established vehicle dynamics model.

9. The adaptive vehicle tracking control method based on a main loop-servo loop layered architecture according to claim 7, wherein: The linear tire model F y =C α α, C α is the tire cornering stiffness, α=(v+l F r) / V-δ F is the front axle slip angle, where v is the lateral speed of the vehicle along the y-axis, V represents the vehicle speed, and δ F is the vehicle's front wheel turning angle.

10. The adaptive vehicle tracking control method based on a main loop-servo loop layered architecture according to claim 6, wherein: The ideal vehicle trajectory tracking control is achieved by using a control algorithm based on geometric relationships or dynamics, or a model-free control algorithm.

Images