Automatic parking reference trajectory optimization method

Abstract

The invention provides an automatic parking reference trajectory optimization method. The automatic parking reference trajectory optimization method comprises the following steps: a, acquiring environment sensing data including a vehicle position, a parking space geometrical shape, obstacle information, a road condition and the like, the data being input information of path planning, performing preliminary path planning based on the acquired environment data, and generating a preliminary reference trajectory, according to the preliminary path planning, a preliminary trajectory Xinit is generated through a global search algorithm, the feasibility of the trajectory is calculated, and the relative position of a key node on a path and an obstacle is determined. According to the automatic parking reference trajectory optimization method, global path planning and local path correction are combined, parking space geometrical morphology, obstacle positions and dynamic environment factors are fully considered, the generated reference trajectory is optimized, and smoothness and feasibility of the path are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of automatic parking, and in particular to an automatic parking reference trajectory optimization method. Background Art

[0002] The method mainly includes three core modules: the environmental perception and path planning module obtains information about the vehicle, surrounding obstacles, parking space geometry, etc. in real time through sensors such as laser radar and cameras, and uses these perception data for preliminary path planning to generate a reference trajectory from the current position of the vehicle to the parking space. The trajectory optimization module optimizes the reference trajectory and uses algorithms such as dynamic programming and particle swarm optimization to smooth and correct the trajectory to ensure that the trajectory will not collide with obstacles and has good feasibility and comfort. Based on the optimized trajectory, the control and execution module calculates and executes steering, acceleration and braking instructions, and transmits these instructions to the vehicle's drive system to ensure that the vehicle can enter the parking space smoothly and safely. The working principle of the system is to generate a safe trajectory that complies with traffic rules by integrating real-time environmental perception, path planning and trajectory optimization technology. The system first perceives the environment and generates a preliminary reference trajectory based on these data, and then corrects the trajectory through the optimization algorithm to improve the smoothness and executability of the trajectory, ultimately ensuring that the vehicle can accurately execute control instructions and enter the warehouse smoothly.

[0003] The main defects of this system include the following points. Trajectory optimization limitations are also one of the defects of this system. Especially in some complex parking scenarios, such as narrow parking spaces or areas with dense obstacles, the optimization algorithm may not be able to generate the optimal path, resulting in an unsmooth or difficult entry process. The system is overly dependent on the vehicle dynamics model. If there is a deviation between the model and the actual vehicle behavior, such as factors such as road friction or speed changes, it may cause errors in the control instructions, causing the vehicle to deviate when executing the trajectory, affecting the stability and reliability of the system. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention provides a method for optimizing the reference trajectory of automatic parking, which solves the problem that the optimization algorithm may not be able to generate the optimal path in narrow parking spaces or areas with dense obstacles, resulting in an unstable parking process or difficulty in execution; factors such as road friction or vehicle speed changes may cause errors in control commands, causing deviations in vehicle trajectory execution, affecting the stability and reliability of the system.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: an automatic parking reference trajectory optimization method, comprising the following steps:

[0006] a. Obtain environmental perception data, including vehicle position, parking space geometry, obstacle information, and road conditions. The data is the input information for path planning and is defined as the environmental state data set D env ,in:

[0007] D env = {P car , P slot , P obstacle , μ road}

[0008] Among them, P car represents the vehicle position, P slot Indicates the parking space position, P obstacle represents the obstacle position, μ road is the road friction coefficient;

[0009] b. Perform preliminary path planning based on the acquired environmental data to generate a preliminary reference trajectory. The preliminary path planning generates a preliminary trajectory X through a global search algorithm. init , and calculate the feasibility of the trajectory, determine the relative positions of key nodes and obstacles on the path, and the global path optimization model can be expressed as:

[0010] X init =GlobalPath(D env )

[0011] Among them, GlobalPath is a global path planning algorithm;

[0012] c. Optimize the preliminary reference trajectory, use multiple optimization algorithms to smooth, adjust and correct the trajectory to generate an optimized entry trajectory. The optimization goal is to minimize the turning radius and the total length of the path. The optimization objective function is:

[0013]

[0014] Where Δx i and Δy i is the lateral and longitudinal deviation of the trajectory point, θ i is the steering angle of the trajectory, λ1, λ2 are the adjustment weights;

[0015] d. Introduce a dynamic feedback mechanism to update factors such as vehicle speed, steering angle, and road friction coefficient in real time, and make trajectory corrections based on real-time environmental data. The dynamic feedback correction can be described by the following model:

[0016] X corrected =X opt +ΔX dyn

[0017] Where ΔXdyn Trajectory correction based on real-time dynamic data (such as vehicle speed and friction coefficient);

[0018] e. Based on the optimized entry trajectory, calculate and generate control instructions, including steering angle, acceleration and braking instructions. The control instruction model is:

[0019] u k =K steer ·θ k +K speed ·v k

[0020] Among them, θ k is the steering angle of the kth step, v k is the vehicle speed, K steer and K speed is the control gain for steering and speed;

[0021] f. Transmit the control command to the vehicle's drive system, execute the vehicle entry operation, and dynamically adjust the control command based on real-time sensor feedback. The control command update can be expressed as:

[0022] u new =u current +Δu feedback

[0023] Among them, Δu feedback is the feedback correction amount;

[0024] g. During the execution process, the relative positions of vehicles, obstacles and parking spaces are monitored in real time, and the control parameters during trajectory execution are dynamically adjusted to ensure smooth and stable path execution. The real-time monitoring model is:

[0025] P relative =(P car -P obstacle )

[0026] And ensure that the vehicle avoids collision with obstacles by adjusting control parameters;

[0027] h. If there is a deviation during the vehicle's execution, the feedback correction algorithm is used to partially correct the trajectory to ensure that the vehicle enters the warehouse smoothly and steadily. The deviation correction model is:

[0028] X corrected =X opt +f(ΔP error , Δu)

[0029] Where f is the feedback correction function, ΔP error is the deviation, and Δu is the corrected control instruction.

[0030] Preferably, the optimization algorithm includes a global path optimization algorithm based on dynamic programming and a local path correction algorithm based on particle swarm optimization, and the dynamic feedback mechanism can adjust the control parameters of the vehicle in real time in response to dynamic factors such as vehicle speed, friction coefficient, steering angle, etc.

[0031] Preferably, the control instructions include steering angle, braking and acceleration control instructions for front and rear wheels, and vehicle speed adjustment instructions, and the path planning algorithm is dynamically switched and adjusted according to different types of parking scenarios.

[0032] Preferably, when processing dynamic obstacles, the optimization algorithm adopts a model based on time series prediction to predict the movement trajectory of the obstacle in advance and avoid potential collision risks in path planning.

[0033] Preferably, the vehicle dynamic model includes the vehicle's dynamic characteristics, a steering system model and a speed response model. The vehicle dynamic model combines the current vehicle speed and steering angle to perform real-time prediction and accurately correct the trajectory execution.

[0034] Preferably, the vehicle includes a precise execution module based on a sliding mode control algorithm to cope with the impact of external disturbances and system uncertainties on vehicle execution.

[0035] Preferably, the feedback correction algorithm uses a fuzzy controller to dynamically adjust the control instructions of the vehicle speed and steering angle to cope with the real-time changes in the road friction coefficient and the dynamic position of the obstacle.

[0036] Preferably, the local path correction algorithm adopts a combination of global and local searches based on a genetic algorithm to further optimize the turning angles in the path and the smoothness of the vehicle trajectory.

[0037] The present invention provides a method for optimizing reference trajectory of automatic parking. It has the following beneficial effects:

[0038] The automatic parking reference trajectory optimization method combines global path planning with local path correction, fully considers the parking space geometry, obstacle location and dynamic environmental factors, optimizes the generated reference trajectory, and ensures the smoothness and feasibility of the path. In addition, the dynamic feedback mechanism makes corrections based on factors such as real-time vehicle speed, steering angle, and road friction coefficient, and can adjust the control parameters in path execution in real time to avoid deviations during path execution, thereby ensuring that the vehicle can smoothly complete parking.

[0039] The present invention further improves the stability and reliability of the system. By introducing precise execution modules based on sliding mode control algorithms, dynamic adjustment of control instructions by fuzzy controllers, and obstacle trajectory prediction based on time series prediction, it can effectively cope with external disturbances and system uncertainties, and ensure the stability and accuracy of vehicle execution in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 Schematic diagram showing the comparison of optimization effects of the reference trajectory for automatic parking in the present invention.

[0041] Figure 2 This is a schematic diagram comparing the turning angles of the paths before and after optimization of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] like Figure 1-2 As shown, an embodiment of the present invention provides an automatic parking reference trajectory optimization method, including: a. obtaining environmental perception data, including vehicle position, parking space geometry, obstacle information, and road conditions, etc. The data is input information for path planning and is defined as an environmental state data set D env ,in:

[0044] D env = {P car , P slot , P obstacle , μ road}

[0045] Among them, P car represents the vehicle position, P slot Indicates the parking space position, P obstacle represents the obstacle position, μ road is the road friction coefficient.

[0046] b. Perform preliminary path planning based on the acquired environmental data to generate a preliminary reference trajectory. The preliminary path planning generates a preliminary trajectory X through a global search algorithm. init , and calculate the feasibility of the trajectory, determine the relative positions of key nodes and obstacles on the path, and the global path optimization model can be expressed as:

[0047] X init =GlobalPath(D env)

[0048] Among them, GlobalPath is a global path planning algorithm. In the preliminary path planning, the vehicle's dynamic model and environmental perception data are combined to consider the vehicle's turning radius, maximum acceleration, maximum braking torque and other physical constraints to ensure that the generated preliminary path meets the vehicle's physical capability limitations. The vehicle's dynamic constraints can be described by the following model:

[0049] F acc =ma max , F friction =μmg

[0050] in, Indicates vehicle acceleration, F acc is the vehicle acceleration force, F friction is the road friction, μ is the static friction coefficient, m is the vehicle mass, a max is the maximum acceleration and g is the acceleration due to gravity.

[0051] c. Optimize the preliminary reference trajectory and use multiple optimization algorithms to smooth, adjust and correct the trajectory to generate an optimized entry trajectory. The optimization goal is to minimize the turning radius and total length of the path. The optimization objective function is:

[0052]

[0053] Where Δx i and Δy i is the lateral and longitudinal deviation of the trajectory point, θ i is the steering angle of the trajectory, λ1 and λ2 are adjustment weights. The optimization algorithm includes a global path optimization algorithm based on dynamic programming and a local path correction algorithm based on particle swarm optimization. The dynamic feedback mechanism can adjust the control parameters of the vehicle in real time and respond to dynamic factors such as vehicle speed, friction coefficient, and steering angle. When dealing with dynamic obstacles, the optimization algorithm adopts a model based on time series prediction to predict the movement trajectory of obstacles in advance and avoid potential collision risks in path planning. The local path correction algorithm adopts a combination of global and local searches based on genetic algorithms to further optimize the turning angle in the path and the smoothness of the vehicle trajectory. The optimization algorithm adopts a multi-stage optimization strategy based on dynamic programming and particle swarm optimization. It not only optimizes in the path generation stage, but also calculates the steering angle and vehicle speed in real time during the trajectory adjustment process of each stage to ensure the smoothness of the path and reduce violent steering. The optimization goal also includes the dynamic response balance of the vehicle, considering the nonlinear relationship between vehicle steering and vehicle speed.

[0054] d. Introduce a dynamic feedback mechanism to update factors such as vehicle speed, steering angle, and road friction coefficient in real time, and make trajectory corrections based on real-time environmental data. The dynamic feedback correction can be described by the following model:

[0055] X corrected =X opt +ΔX dyn

[0056] Where ΔX dyn In order to correct the trajectory based on real-time dynamic data, the dynamic feedback mechanism not only takes into account the vehicle speed, friction coefficient and steering angle, but also introduces external disturbance factors such as wind force, road wetness, etc. These factors are corrected in real time based on sensor data and prediction models to further improve control accuracy.

[0057] e. Based on the optimized entry trajectory, calculate and generate control instructions, including steering angle, acceleration and braking instructions. The control instruction model is:

[0058] u k =K steer ·θ k +K speed ·v k

[0059] Among them, θ k is the steering angle of the kth step, v k is the vehicle speed, K steer and K speed The control gains for steering and speed include steering angle, braking and acceleration control instructions for the front and rear wheels, and vehicle speed adjustment instructions. The path planning algorithm dynamically switches and adjusts according to different types of parking scenarios.

[0060] f. Transmit the control command to the vehicle's drive system, execute the vehicle entry operation, and dynamically adjust the control command based on real-time sensor feedback. The control command update can be expressed as:

[0061] u new =u current +Δu feedback

[0062] Among them, Δu feedback is the feedback correction amount.

[0063] g. During the execution process, the relative positions of vehicles, obstacles and parking spaces are monitored in real time, and the control parameters during trajectory execution are dynamically adjusted to ensure smooth and stable path execution. The real-time monitoring model is:

[0064] P relative =(P car -P obstacle )

[0065] And ensure that the vehicle avoids collision with obstacles by adjusting control parameters.

[0066] h. If there is a deviation during the vehicle's execution, the feedback correction algorithm is used to partially correct the trajectory to ensure that the vehicle enters the warehouse smoothly and steadily. The deviation correction model is:

[0067] X corrected =X opt +f(ΔP error , Δu)

[0068] Where f is the feedback correction function, ΔP error is the deviation, Δu is the corrected control instruction, the dynamic model of the vehicle includes the vehicle's dynamic characteristics, steering system model and speed response model. The vehicle dynamic model combines the current vehicle speed and steering angle for real-time prediction, and accurately corrects the trajectory execution. The vehicle includes a precise execution module based on the sliding mode control algorithm to cope with the impact of external disturbances and system uncertainty on vehicle execution. The feedback correction algorithm uses a fuzzy controller to dynamically adjust the control instructions of the vehicle speed and steering angle to cope with real-time changes in road friction coefficient and the dynamic position of obstacles.

[0069] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing reference trajectory of automatic parking, characterized in that: The following steps are involved: a. Obtain environmental perception data, including vehicle position, parking space geometry, obstacle information, and road conditions. The data is the input information for path planning and is defined as the environmental state data set D env ,in: D env ={P car ,P slot ,P obstacle ,μ road } Among them, P car represents the vehicle position, P slot Indicates the parking space position, P obstacle represents the obstacle position, μ road is the road friction coefficient; b. Perform preliminary path planning based on the acquired environmental data to generate a preliminary reference trajectory. The preliminary path planning generates a preliminary trajectory X through a global search algorithm. init , and calculate the feasibility of the trajectory, determine the relative positions of the key nodes and obstacles on the path, and the global path optimization model can be expressed as: X init =GlobalPath(D env ) Among them, GlobalPath is a global path planning algorithm; c. Optimize the preliminary reference trajectory, use multiple optimization algorithms to smooth, adjust and correct the trajectory to generate an optimized entry trajectory. The optimization goal is to minimize the turning radius and the total length of the path. The optimization objective function is: Where Δx i and Δy i is the lateral and longitudinal deviation of the trajectory point, θ i is the steering angle of the trajectory, λ1, λ2 are the adjustment weights; d. Introduce a dynamic feedback mechanism to update vehicle speed, steering angle, road friction coefficient and other factors in real time, and make trajectory corrections based on real-time environmental data. The dynamic feedback correction can be described by the following model: X corrected =X opt +ΔX dyn Where ΔX dyn Trajectory correction based on real-time dynamic data (such as vehicle speed and friction coefficient); e. Based on the optimized entry trajectory, calculate and generate control instructions, including steering angle, acceleration and braking instructions. The control instruction model is: u k =K steer ·θ k +K speed ·v k Among them, θ k is the steering angle of the kth step, v k is the vehicle speed, K steer and K speed is the control gain for steering and speed; f. Transmit the control command to the vehicle's drive system, execute the vehicle entry operation, and dynamically adjust the control command based on real-time sensor feedback. The control command update can be expressed as: u new =u current +Δu feedback Among them, Δu feedback is the feedback correction amount; g. During the execution process, the relative positions of vehicles, obstacles and parking spaces are monitored in real time, and the control parameters during trajectory execution are dynamically adjusted to ensure smooth and stable path execution. The real-time monitoring model is: P relative =(P car -P obstacle ) And ensure that the vehicle avoids collision with obstacles by adjusting control parameters; h. If there is a deviation during the vehicle's execution, the feedback correction algorithm is used to partially correct the trajectory to ensure that the vehicle enters the warehouse smoothly and steadily. The deviation correction model is: X corrected =X opt +f(ΔP error ,Δu) Where f is the feedback correction function, ΔP error is the deviation, and Δu is the corrected control instruction.

2. The method for optimizing reference trajectory of automatic parking according to claim 1, characterized in that: The optimization algorithm includes a global path optimization algorithm based on dynamic programming and a local path correction algorithm based on particle swarm optimization. The dynamic feedback mechanism can adjust the control parameters of the vehicle in real time and respond to dynamic factors such as vehicle speed, friction coefficient, steering angle, etc.

3. The method for optimizing reference trajectory of automatic parking according to claim 1, characterized in that: The control instructions include steering angle, braking and acceleration control instructions for front and rear wheels, and vehicle speed adjustment instructions. The path planning algorithm is dynamically switched and adjusted according to different types of parking scenarios.

4. The method for optimizing reference trajectory of automatic parking according to claim 1, characterized in that: When dealing with dynamic obstacles, the optimization algorithm adopts a model based on time series prediction to predict the movement trajectory of the obstacle in advance and avoid potential collision risks in path planning.

5. The method for optimizing reference trajectory of automatic parking according to claim 1, characterized in that: The vehicle dynamic model includes the vehicle's dynamic characteristics, steering system model and speed response model. The vehicle dynamic model combines the current vehicle speed and steering angle to make real-time predictions and accurately correct trajectory execution.

6. The method for optimizing reference trajectory of automatic parking according to claim 5, characterized in that: The vehicle includes a precise execution module based on a sliding mode control algorithm to cope with the impact of external disturbances and system uncertainties on vehicle execution.

7. The method for optimizing reference trajectory of automatic parking according to claim 1, characterized in that: The feedback correction algorithm uses a fuzzy controller to dynamically adjust the control instructions of vehicle speed and steering angle to cope with real-time changes in road friction coefficient and the dynamic position of obstacles.

8. The method for optimizing reference trajectory of automatic parking according to claim 2, characterized in that: The local path correction algorithm adopts a combination of global and local searches based on a genetic algorithm to further optimize the turning angles in the path and the smoothness of the vehicle trajectory.

Images