Automatic parking control method and system based on large model and storage medium

Abstract

The invention relates to an automatic parking control method and system based on a large model and a storage medium, and the method comprises the following steps: carrying out data initialization, obtaining and processing a parking lot map, and generating a long-term reference trajectory; monitoring the environment where the autonomous vehicle is located based on a large model technology and predicting the movement of an obstacle; checking whether the autonomous vehicle is safe or not according to the predicted obstacle movement result, and if yes, performing obstacle avoidance planning; if not, whether the long-term reference track needs to be repaired due to movement of the obstacle or not is checked, and if yes, the long-term reference track is optimized to obtain a travelable self-vehicle movement track; otherwise, the latest reference trajectory in the dynamic environment is tracked based on model predictive control. Compared with the prior art, the identification and judgment capability of the automatic driving technology under the complex road condition is improved, and the safety and the automation level of automatic driving are improved on the whole.

Description

Technical Field

[0001] The present invention relates to a field, and in particular to an automatic parking control method, system and storage medium based on a large model. Background Art

[0002] Fully automated parking in the existing technology remains challenging, especially in a dynamic environment with multiple independent agents, not only because it involves motion planning in a small space, but also because the autonomous vehicle (AV) should react intelligently to the surrounding obstacle vehicles (OV). Compared with driving on roads or highways, there is no clear set of rules to follow for vehicle movement in parking areas, which depends largely on the driver's intention and even skill level. These make the parking environment prediction quite challenging, so in order to have a good level of control over the entire automated parking control, an autonomous parking system that integrates prediction and planning is also required.

[0003] In the process of automatic parking, motion prediction is crucial because it determines the safety constraints of the planning module, thereby determining the feasibility and smoothness of motion planning. The existing automatic parking process mainly has the following technical problems:

[0004] 1) Autonomous driving technology cannot accurately identify and judge under complex road conditions; 2) In the case of large-scale vehicle access, Internet of Vehicles technology will encounter network congestion and data transmission delay problems; 3) Automatic parking systems cannot accurately identify and judge in narrow parking spaces or complex parking environments. Summary of the invention

[0005] The purpose of the present invention is to provide an automatic parking control method, system and storage medium based on a large model in order to overcome the defect of the above-mentioned prior art that the autonomous driving technology cannot accurately identify and judge under complex road conditions.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] An automatic parking control method based on a large model comprises the following steps:

[0008] S1: Initialize data, obtain and process the parking lot map, and generate a long-term reference trajectory;

[0009] S2: Monitor the environment of the autonomous vehicle and predict the movement of obstacles based on large model technology;

[0010] S3: Check whether the autonomous driving vehicle is safe based on the predicted obstacle movement result. If so, perform obstacle avoidance planning; if not, execute step S4;

[0011] S4: Check whether the long-term reference trajectory needs to be repaired due to the movement of the obstacle. If necessary, execute step S5; otherwise, execute step S6;

[0012] S5: Optimizing the long-term reference trajectory to obtain a drivable vehicle motion trajectory; if optimization is not possible, updating the parking lot map to regenerate the long-term reference trajectory;

[0013] S6: Tracking the latest reference trajectory in a dynamic environment based on model predictive control.

[0014] Further, the obstacle movement prediction process includes a short-term motion prediction process and a long-term pattern prediction process;

[0015] The short-term motion prediction process includes determining the corresponding forward propagation estimation state according to the state average value of the obstacle, thereby obtaining the short-term motion prediction result of the obstacle;

[0016] The long-term mode prediction process includes determining the long-term motion mode of the obstacle based on the historical state, dynamic model and movement of the obstacle relative to the environment of the obstacle; the historical state and dynamic model of the obstacle are obtained through the short-term motion prediction process, and the movement of the obstacle relative to the environment is used to combine with the cost map to determine the motion mode of the obstacle, form a safety margin and a safety boundary, and finally determine the long-term motion prediction result of the obstacle. The safety margin is used to determine whether the autonomous driving vehicle is safe, and the safety boundary is used to limit the movement trajectory of the autonomous driving vehicle.

[0017] Furthermore, the model predictive control process is used to track a reference trajectory with a given safety margin and a safety margin, and the tracking trajectory is calculated by optimizing the basic planner.

[0018] Furthermore, the obstacle avoidance planning is used to be executed when the movement of the obstacle violates the corresponding safety margin, and to find the best target by exploring the forward drivable environment.

[0019] Furthermore, when the reference trajectory needs to be repaired due to the movement of an obstacle, if the area ahead is not feasible, the autonomous driving vehicle is controlled to stop on the reference trajectory and wait for the obstacle to disappear; if the waiting time exceeds a preset waiting threshold, the reference trajectory is updated to guide the autonomous driving vehicle to bypass the obstacle and return to the original path.

[0020] Furthermore, in step S2, big data and deep learning technology are used to monitor environmental information around the vehicle.

[0021] Furthermore, the method realizes real-time interaction and information sharing between the autonomous driving vehicle and the surrounding environment through collaborative communication among the autonomous driving vehicle, the roadside terminal and the Internet of Vehicles in the cloud.

[0022] Furthermore, the Internet of Vehicles collaborative communication adopts a decentralized network structure, and data communication and processing are carried out by multi-node parallel processing and load balancing technology.

[0023] The present invention also provides an automatic parking control system based on a large model, comprising a memory and a processor, wherein the memory stores a computer program, and the processor calls the computer program to execute the steps of the method described above.

[0024] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and the computer program is executed by a processor to perform the above method.

[0025] Compared with the prior art, the present invention has the following advantages:

[0026] (1) The present invention adopts an automatic parking collaborative technology solution based on a large model, which can more accurately identify and judge the environmental information around the vehicle under complex road conditions through big data and deep learning technology; further realize the prediction of short-term and long-term movements of obstacles; according to the prediction results, it first detects whether the autonomous driving vehicle violates the safety margin through motion planning; if not, it checks whether the initial long-term reference trajectory needs to be repaired due to the movement of the obstacle; if any of these situations occurs, the reference trajectory will be updated; finally, a collision-free motion is planned based on the model predictive controller to track the latest long-term motion trajectory in the dynamic environment; the overall safety and automation level of autonomous driving are improved.

[0027] (2) Optimizing the performance of the Internet of Vehicles: The present invention adopts an automatic parking collaborative technical solution based on a large model. It can effectively solve the problems of network congestion and data transmission delay through distributed computing and edge computing technologies when large-scale vehicles are connected, thereby optimizing the performance of the Internet of Vehicles and improving the efficiency of information sharing and exchange.

[0028] (3) Improving the convenience of automatic parking: The present invention adopts an automatic parking collaborative technical solution based on a large model. It can generate a better parking path through a computer algorithm in a narrow parking space or complex parking environment, thereby improving the convenience of automatic parking and reducing the complexity and difficulty of the driver's operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 A schematic flow chart of an automatic parking control method based on a large model provided in an embodiment of the present invention;

[0030] Figure 2 The figure is a schematic diagram of a framework flow of an automatic parking control method based on a large model provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0032] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. 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.

[0033] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0034] Example 1

[0035] Typical automatic parking planning control is introduced as follows:

[0036] Automatic parking motion planning for intelligent vehicles is a major challenge in automatic parking scenarios. General motion planning algorithms cannot be directly applied to parking in places with obstacles, as this requires rapid replanning of complex driving operations. On the other hand, motion planning specialized in automatic parking either fails to combine short-term planning with long-term planning or fails to incorporate online path repair into new obstacles in dynamic environments. Scenario-aware planners that implement multiple strategies may be efficient in terms of computational time, resulting in a high replanning rate to ensure safety. In general, the regulation and control module of automatic parking mainly consists of four parts: obstacle vehicle trajectory prediction, drivable area selection, local trajectory planning, and vehicle control.

[0037] In the typical software distribution mode of parking path planning, the prediction can use the Interacting multiple model (IMM) filter, input the position of the vehicle, the heading angle, velocity, acceleration, and yaw-rate of the obstacle vehicle, and the output intention is divided into two types: go straight and avoid oncoming vehicles. The "drivable area" chooses to use the potential field method. The obstacle prediction trajectory, road edge, and the position of static obstacles can all be used as the input of the drivable boundary, thereby constructing a potential energy function to evaluate the danger and whether it is drivable. The "local path planning" can use the RRT algorithm, which uses two tricks to increase the speed: divide the drivable area into seeding surfaces and randomly sample on these surfaces; use the inverse of the potential energy function as the probability density function of sampling to reduce the number of sampling points. Find the minimum cost path among the sampling points as a dynamic programming (DP) method. The output trajectory can be smoothed using B-spline (piecewise polynomial).

[0038] like Figure 1 As shown, this embodiment provides an automatic parking control method based on a large model, comprising the following steps:

[0039] S1: Initialize data, obtain and process the parking lot map, and generate a long-term reference trajectory;

[0040] S2: Monitor the environment of the autonomous vehicle and predict the movement of obstacles based on large model technology;

[0041] Specifically, big data and deep learning technologies are used to monitor environmental information around the vehicle;

[0042] The obstacle movement prediction process includes short-term motion prediction process and long-term pattern prediction process;

[0043] The short-term motion prediction process includes determining the corresponding forward propagation estimation state according to the state average of the obstacle, thereby obtaining the short-term motion prediction result of the obstacle;

[0044] The long-term pattern prediction process includes determining the long-term motion pattern of the obstacle based on the historical state, dynamic model and movement of the obstacle relative to the environment. The historical state and dynamic model of the obstacle are obtained through the short-term motion prediction process. The movement of the obstacle relative to the environment is used to combine the cost map to determine the movement pattern of the obstacle, form a safety margin and a safety boundary, and finally determine the long-term motion prediction result of the obstacle. The safety margin is used to determine whether the autonomous driving vehicle is safe, and the safety boundary is used to limit the movement trajectory of the autonomous driving vehicle.

[0045] S3: Check whether the autonomous driving vehicle is safe based on the predicted obstacle movement result. If so, perform obstacle avoidance planning; if not, execute step S4;

[0046] Specifically, obstacle avoidance planning is used to execute when the motion of an obstacle violates the corresponding safety margin, and to find the optimal target by exploring the forward drivable environment;

[0047] S4: Check whether the long-term reference trajectory needs to be repaired due to the movement of the obstacle. If necessary, execute step S5; otherwise, execute step S6;

[0048] Specifically, when the reference trajectory needs to be repaired due to the movement of an obstacle, if the area ahead is not feasible, the autonomous driving vehicle is controlled to stop on the reference trajectory and wait for the obstacle to disappear; if the waiting time exceeds the preset waiting threshold, the reference trajectory is updated to guide the autonomous driving vehicle to bypass the obstacle and return to the original path.

[0049] S5: Optimize the long-term reference trajectory to obtain a drivable ego-vehicle motion trajectory; if optimization is not possible, update the parking lot map and regenerate the long-term reference trajectory;

[0050] S6: Tracking the latest reference trajectory in a dynamic environment based on model predictive control;

[0051] The model predictive control process is used to track the reference trajectory with a given safety margin and safety margin, and the tracking trajectory is calculated by optimizing the basic planner.

[0052] This scheme optimizes the effective information based on the basic parking algorithm. First, a long-term motion reference is generated using a bidirectional A-search guide tree (BIAGT). Then, a strategic motion planner based on the hybrid environment predictor results implements three strategies: a safety controller based on model predictive control (MPC) for trajectory tracking if the reference is still valid for the environment; search-based collision avoidance path planning to quickly find a collision avoidance path in an emergency; and optimization-based repair available path planning when the reference fails. Based on the above operations, the model-based hybrid environment predictor can be used to plan the predicted short-term and long-term motion patterns mentioned above. As a result, a strategic motion planner is developed that can achieve effective planning in different situations.

[0053] The specific process is as follows:

[0054] For the system-level strategy of automatic parking, there are currently interactive multi-model IMMs used for prediction in automatic parking systems in static environments, and sampling-based methods are used for planning. This method first predicts obstacle information and then selects the navigation strategy of the autonomous vehicle. For intelligent vehicles, it is necessary to cross the road in the parking lot, but no parking operation is performed. In contrast, after adopting a more complete integrated parking system approach, short-term and long-term predictions of the environment can be achieved, which can be used for autonomous parking strategy planning.

[0055] A. Motion Planning Definition

[0056] The planning problem considering vehicle dynamics can be expressed as: X' = f(X) + g(X;u); (1)

[0057] Where X = [x; y; θ]T represents the two-dimensional coordinates and vehicle heading, and u = [δ; v]T is the control input including longitudinal speed and steering angle.

[0058] Collision-free configuration space is the configuration set where the vehicle has no intersection with obstacles. The motion planning problem considered in this paper can be defined as given the initial configuration X0 free, the target configuration Xf Cfree and formula (1), find a feasible trajectory Pt, which starts from 0 and ends at Xf, and satisfies (1) and (2) in the collision-free configuration space Cfree. The common bicycle model is used to represent vehicle motion. The discrete time model obtained by Euler discretization is as follows:

[0059] (2)Where Ts is the sampling time.

[0060] B. System Architecture

[0061] like Figure 2 Figure 2 shows the proposed automatic parking system architecture. The hybrid environment predictor and the strategic motion planner are the two main components. At runtime, the central control first processes the parking map Mmap and generates an initial long-term trajectory Pref. Here, the search guide tree BIAGT is used to plan and generate a trajectory that allows the autonomous vehicle to accurately reach the target, which is an important feature for parking in a narrow space.

[0062] The hybrid environment predictor monitors the environment and predicts the movement of obstacles. Based on the prediction, the strategic motion planner checks whether the autonomous vehicle violates the safety margin. If not, it checks whether the initial long-term predicted trajectory Pref needs to be repaired due to the movement of obstacles. If any of these situations occur, Pref will be updated. Finally, a collision-free motion is planned based on the model predictive controller MPC to track the latest long-term motion trajectory Pref in the dynamic environment. If the repair planner cannot successfully optimize a drivable self-vehicle motion trajectory, it will request the central domain control to update the map and regenerate the reference trajectory.

[0063] C. Hybrid Environment Predictor

[0064] The hybrid environment predictor contains three main parts: obstacle motion estimation, motion prediction, and pattern estimation. In particular, accurate short-term motion prediction enables intelligent vehicles to plan and react safely to obstacles; a major challenge in short-term prediction is to estimate the steering angle of obstacles. An extended Kalman filter (EKF) can usually be used to reconstruct the speed of the obstacle, and then an adaptive observer is used for steering estimation. Long-term planning / pattern prediction can ensure that intelligent vehicles plan reasonable driving paths more efficiently and smoothly during operation. Although long-term motion is usually difficult to predict accurately, it can be observed that drivers usually follow some routes due to driving habits (for example, cars should drive on the right in China). In addition, the movement of the vehicle throughout the parking / leaving process can be captured by several "patterns" (for example, maneuvering in and out of narrow spaces and cruising in aisles). Based on these two priors, cost maps can be used to capture these routes, combined with short-term predictions to determine the pattern of obstacles, and make long-term predictions.

[0065] The following is an explanation of its main features.

[0066] 1) Cascade motion estimation:

[0067] Motion estimation reconstructs the state X from the measurements of (x; y) based on the wheelbarrow model. Such processing is insufficient for parking processes involving frequent changes in movement direction and steering actions. In order to accurately predict the short-term motion of the obstacle, it is advantageous to reconstruct the control input u. It can be regarded as an unknown input estimation problem or the system state of the obstacle vehicle OV is augmented with the control input and the state estimation problem is solved. Assume that the obstacle motion evolves according to model (2). The augmented model of OV is obtained by assuming that the control input (δ; v) is a piecewise constant, and the augmented state [x; y; θ; δ; v]T is estimated. Given the nonlinear augmented model, it is natural to apply a well-established nonlinear state estimator (such as an extended Kalman filter EKF or a particle filter) for state estimation. It is not simple to accurately estimate the steering angle by adjusting the EKF, which is partly due to the existence of the term v tan(δ) involving the multiplication of the unmeasured state. Although particle filters are good, their heavy computation may also hinder practical operations.

[0068] 2) Short-term motion prediction:

[0069] Short-term motion prediction focuses more on computational efficiency and real-time performance. We assume that the short-term motion of the obstacle vehicle OV is completely captured by the average value of the state Xcc and its covariance. The state is estimated by forward propagation as Xcc, k and the short-term prediction XH is obtained; k = [X1, kT, ..., XH, kT]T is used for the future H steps of the time range. Similarly, forward propagation is performed according to the forward prediction formula of the extended Kalman filter EKF to obtain the covariance matrix PmH, k = {Pmk+1, ..., Pmk+H}, which will help long-term prediction and be used to determine the safety margin for each future time step.

[0070] 3) Long-term pattern prediction:

[0071] The movement of the obstacle vehicle OV over a period of time, which is of interest in long-term motion, depends on its historical state, dynamic model, and its motion relative to the environment. The first two factors are captured to some extent by short-term motion prediction. In order to exploit relative environmental motion, a cost map can be introduced (as shown in the figure below, the hybrid predictor predicts the short-term OV trajectory (green line) and combines it with the pattern prediction to generate safety margins and safety boundaries for h=1 and h=H.) to capture the possible long-term motion of the obstacle vehicle OV.

[0072] At the same time, a route planner can also be used to build a cost map path Mroute, where the cost map contains possible routes taken by the obstacle vehicle OV.

[0073] Where Xmk;i is the i-th waypoint of the route in the pattern;, W1 and W2 are weight matrices. The function f(mk;Xcc;k) is proportional to the size of the OV steering angle and the deviation of the OV heading angle from the final heading angle of the route. Finally, normalize to obtain b(mk), and take the value of mk with the maximum confidence as m^k. In addition, vehicles in parking lots usually operate in two modes, "manual mode" and "cruise mode". Vehicles in manual mode are likely to frequently change steering and deviate from the route in the cost map (the black dashed line in the figure above) to park or leave a narrow parking space. Vehicles in cruise mode have a small or stable steering angle and usually follow one of the routes. The vehicle is in this mode when it first enters the parking lot and approaches a parking space or when the vehicle leaves a parking space and leaves the parking lot.

[0074] 4) Safety margin and safety limit:

[0075] Since the movement of the obstacle vehicle OV in the maneuvering mode is difficult to predict, the predictor needs to generate a safety margin for the ego vehicle's movement (the convex boundary of the OV's historical posture, the orange line in the figure above), which causes the planning model to be more conservative and ensure that the ego vehicle is kept away from the unpredictable obstacle vehicle OV. Note that given the kinematic model and route information, the safety margin and safety margin can also be applied to other moving obstacles, such as pedestrians or motorcycles.

[0076] 5) Strategic Motion Planner

[0077] Using the reference trajectory Pref, the main modules of the motion strategy planner include the MPC-based safety controller and two supporting modules: the collision avoidance trajectory planner and the repair trajectory planner. If the current position of the ego vehicle and the reference trajectory are invalid due to the motion of the ego vehicle, both planners will be activated.

[0078] A. MPC-based safety controller: The safety controller tracks the reference trajectory Pref given a safety margin and a safety margin. With the help of Pref, the tracking trajectory in the MPC framework is calculated using an optimization-based planner. Assume Xref;k is a fragment of Pref and set the time step k for tracking. The external reference k will be selected and pruned to ensure that the required safety margin in all modes or the safety margin in the "maneuver" mode is not violated. The entire trajectory tracking problem can be formulated as a solution to a non-convex optimization problem, which can be solved using a nonlinear programming solver (such as IPOPT) with the reference path as a warm start, and the average solution time is about 0:06 seconds.

[0079] B. Obstacle Avoidance Planning: When the ego vehicle stays on the original reference trajectory, it may be considered unsafe, and the obstacle avoidance planning procedure needs to be initiated. This may happen when the obstacle OV is approaching the ego vehicle, and its motion is very different from the previous prediction - it may violate the safety margin and pose a safety threat. Therefore, the ego vehicle needs to find a way to avoid danger. Backward motion is generally not recommended because the ego vehicle has no safe goal. Instead, exploring the forward drivable environment to find the best goal is optional, so a search-based obstacle avoidance planner is proposed that can explore the space and quickly find an obstacle avoidance trajectory.

[0080] 6) Trajectory Restoration Planner:

[0081] The action of the obstacle vehicle OV may invalidate the reference trajectory of the ego vehicle. When the area ahead is not feasible, the safety controller will command the ego vehicle to stop on the reference trajectory. Unless a new reference trajectory is received, the safety controller will control the ego vehicle to stop moving and wait for the obstacle OV to disappear - if the OV stops for a long time, it is considered that the trajectory planning is unreasonable or inefficient. At this time, the reference trajectory Pref should be updated so that the safety planner can guide the ego vehicle to bypass the OV and merge back to the original path. Note that the repaired trajectory is usually in the same category of trajectory set as the original trajectory, so an executable trajectory can be generated directly by repairing safety hazards based on optimizing the current trajectory. In order to quickly obtain the repair path, it is necessary to repair the trajectory planning in two-dimensional space, that is, set the trajectory to Xrepair = [x; y] T , and modify the constraints accordingly. The resulting path, although collision-free, does not ensure that the tracking is always accurate, which may cause the autonomous vehicle to hit obstacles. Therefore, only when it passes the kinematic feasibility check will it be recognized as a verified path and accepted as the repaired trajectory. If the repair fails, the central control is notified to take over the repair task.

[0082] This embodiment also provides an automatic parking control system based on a large model, including a memory and a processor, wherein the memory stores a computer program, and the processor calls the computer program to execute the steps of the method described above.

[0083] This embodiment further provides a computer-readable storage medium, on which a computer program is stored. The computer program is executed by a processor to perform the above method.

[0084] In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0085] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. An automatic parking control method based on a large model, characterized in that: The following steps are involved: S1: Initialize data, obtain and process the parking lot map, and generate a long-term reference trajectory; S2: Monitor the environment of the autonomous vehicle and predict the movement of obstacles based on large model technology; S3: Check whether the autonomous driving vehicle is safe based on the predicted obstacle movement result. If so, perform obstacle avoidance planning; if not, execute step S4; S4: Check whether the long-term reference trajectory needs to be repaired due to the movement of the obstacle, and if necessary, execute step S5; Otherwise, execute step S6; S5: Optimizing the long-term reference trajectory to obtain a drivable vehicle motion trajectory; if optimization is not possible, updating the parking lot map to regenerate the long-term reference trajectory; S6: Tracking the latest reference trajectory in a dynamic environment based on model predictive control.

2. The automatic parking control method based on a large model according to claim 1, characterized in that: The obstacle movement prediction process includes a short-term motion prediction process and a long-term pattern prediction process; The short-term motion prediction process includes determining the corresponding forward propagation estimation state according to the state average value of the obstacle, thereby obtaining the short-term motion prediction result of the obstacle; The long-term mode prediction process includes determining the long-term motion mode of the obstacle based on the historical state, dynamic model and movement of the obstacle relative to the environment of the obstacle; the historical state and dynamic model of the obstacle are obtained through the short-term motion prediction process, and the movement of the obstacle relative to the environment is used to combine with the cost map to determine the motion mode of the obstacle, form a safety margin and a safety boundary, and finally determine the long-term motion prediction result of the obstacle. The safety margin is used to determine whether the autonomous driving vehicle is safe, and the safety boundary is used to limit the movement trajectory of the autonomous driving vehicle.

3. The large model-based automatic parking control method according to claim 2, characterized in that: The model predictive control process is used to track a reference trajectory with a given safety margin and a safety boundary, and the tracking trajectory is calculated by optimizing the basic planner.

4. The large model-based automatic parking control method according to claim 2, characterized in that: The obstacle avoidance planning is used to be executed when the movement of the obstacle violates the corresponding safety margin, and finds the best target by exploring the forward drivable environment.

5. The large model-based automatic parking control method according to claim 2, characterized in that: When the reference trajectory needs to be repaired due to the movement of an obstacle, if the area ahead is not feasible, the autonomous driving vehicle is controlled to stop on the reference trajectory and wait for the obstacle to disappear; if the waiting time exceeds a preset waiting threshold, the reference trajectory is updated to guide the autonomous driving vehicle to bypass the obstacle and return to the original path.

6. The large model-based automatic parking control method according to claim 1, characterized in that: In step S2, big data and deep learning technology are used to monitor the environmental information around the vehicle.

7. The large model-based automatic parking control method according to claim 1, characterized in that: The method realizes real-time interaction and information sharing between the autonomous driving vehicle and the surrounding environment through collaborative communication among the autonomous driving vehicle, the roadside terminal and the Internet of Vehicles in the cloud.

8. The large model-based automatic parking control method according to claim 7, characterized in that: The Internet of Vehicles collaborative communication adopts a decentralized network structure, and performs data communication and processing by multi-node parallel processing and load balancing technology.

9. An automatic parking control system based on a large model, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and the processor calls the computer program to execute the steps of any one of the methods according to claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is used by a processor to execute the method according to any one of claims 1 to 8.

Images