Logistics park vehicle path planning method and system

Abstract

The invention discloses a logistics park vehicle path planning method and system, and the method comprises the steps: obtaining the real-time position and task state information of a vehicle in a park, dynamically constructing a vehicle path planning network model, dividing a road into nodes and edges, and forming a directed weighted graph; according to the directed weighted graph, an improved multi-lane algorithm is adopted, the path length, the congestion degree and the vehicle characteristic factors are comprehensively considered, and the optimal path of each vehicle is calculated; for the optimal path, introducing a virtual lane and a time dimension, pre-judging occupation conditions of multiple vehicles on the virtual lane for each time slice, and constructing a conflict constraint matrix; obtaining park high-precision map data, extracting road network topology, lane lines and traffic sign marking elements, performing lane-level refinement on the optimal path, and generating a driving track; and through a multi-sensor fusion positioning technology, obtaining high-precision position and attitude information of the vehicle, matching the high-precision position and attitude information with the driving track, and calculating a track tracking error.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle path planning, and in particular relates to a method and system for vehicle path planning in a logistics park. Background Art

[0002] A logistics park refers to a work area where only automatic horizontal handling equipment is used to carry goods horizontally within a limited range. This type of park is designed with spatial resources such as access roads and interactive nodes. Handling equipment is usually allowed to move straight, change lanes, turn, rotate on the spot, etc. on the road. Intersections between roads are allowed, and no traffic lights are set at the intersections. Because road space resources are limited, there are many spatial conflicts between roads in design, especially in some applications, where the roads present a dense network intersection design. The problems faced by vehicle path planning in logistics parks include: (1) Dynamic path change problem: The operation scheduling system of the logistics park often undergoes real-time changes, resulting in real-time and uncertain conflicts between equipment, which puts higher requirements on vehicle path planning; (2) Multi-vehicle shortest path planning problem: It is necessary to consider potential conflicts with other vehicles on future paths, which increases the complexity of path planning; (3) Vehicle safety issues, congestion issues, and task deadlock issues: These issues are the main problems that need to be solved in the operation of closed automated logistics parks, especially when vehicles choose the best path, and they need to be avoided; (4) Spatial conflict issues within the park: Due to limited road space resources, there are many spatial conflicts between roads in design, especially in roads with dense network intersection designs, which poses higher challenges to path planning. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes a logistics park vehicle path planning method and system to improve the efficiency of logistics distribution and reduce transportation costs.

[0004] On the one hand, to achieve the above-mentioned purpose, the present invention provides a method for vehicle path planning in a logistics park, comprising: obtaining real-time position and task status information of vehicles in the park, dynamically constructing a vehicle path planning network model, dividing roads into nodes and edges, and forming a directed weighted graph;

[0005] According to the directed weighted graph, an improved multi-lane algorithm is used to comprehensively consider the path length, congestion level and vehicle characteristics to calculate the optimal path for each vehicle;

[0006] For the optimal path, virtual lanes and time dimensions are introduced, the occupancy of multiple vehicles on the virtual lanes is predicted for each time slice, and a conflict constraint matrix is ​​constructed;

[0007] Obtain high-precision map data of the park, extract road network topology, lane lines, and traffic sign marking elements, perform lane-level refinement on the optimal path, and generate a driving trajectory;

[0008] The high-precision position and posture information of the vehicle is obtained through multi-sensor fusion positioning technology, the high-precision position and posture information of the vehicle is matched with the driving trajectory, and the trajectory tracking error is calculated;

[0009] Vehicles exchange information with other vehicles and roadside facilities through V2X communication technology, and perceive the surrounding environment in real time. If other vehicles or obstacles are detected ahead, collision warning and obstacle avoidance mechanisms are activated.

[0010] According to the logistics park vehicle path planning method provided by the present invention, the real-time position and task status information of the vehicles in the park are obtained, a vehicle path planning network model is dynamically constructed, and the roads are divided into nodes and edges to form a directed weighted graph, including:

[0011] The vehicle positioning device is used to obtain the location coordinates of each vehicle in the park and the status information of the currently executed task in real time, and upload them to the path planning server;

[0012] According to the obtained vehicle position coordinates, the road section where the vehicle is located is marked on the map, the road is divided into several nodes and directed edges connecting the nodes, and a vehicle path planning network model is constructed;

[0013] The constructed network model is analyzed using graph theory algorithms to calculate the weight of each edge. The weight value needs to be set by comprehensively considering factors such as road length and traffic status to form a directed weighted graph for vehicle path planning.

[0014] According to the logistics park vehicle path planning method provided by the present invention, based on the directed weighted graph, an improved multi-lane algorithm is used to comprehensively consider the path length, congestion level, and vehicle characteristics to calculate the optimal path for each vehicle, including:

[0015] According to the structure and weight information of the directed weighted graph, a road network topology model is constructed and stored in a graph database;

[0016] Obtain real-time traffic flow data, dynamically evaluate the degree of road congestion, and associate the congestion index with the road section for storage;

[0017] For the starting point and end point of each vehicle, the improved Dijkstra algorithm is used to calculate multiple alternative routes based on the path length and congestion index;

[0018] According to the characteristic parameters of the vehicle, the adaptability of the alternative routes is scored, and the routes with high matching degree with the vehicle characteristics are selected;

[0019] The alternative paths are further optimized through the ant colony optimization algorithm, which simulates the vehicle interaction behavior in multi-lane scenarios and obtains more reasonable path planning results.

[0020] Match the optimized path with the vehicle information to generate the optimal driving route for each vehicle, and return the result to the vehicle terminal device;

[0021] Continuously monitor changes in road conditions and vehicle positions. If congestion worsens or the vehicle deviates from the planned route, it triggers path replanning to ensure real-time dynamic optimization.

[0022] According to the logistics park vehicle path planning method provided by the present invention, for the optimal path, a virtual lane and a time dimension are introduced, the occupancy of multiple vehicles on the virtual lane is predicted for each time slice, and a conflict constraint matrix is ​​constructed, including:

[0023] According to the optimal path, obtaining virtual lane information and time dimension information on the path;

[0024] For each time slice, obtain the predicted occupancy of multiple vehicles on the virtual lane;

[0025] According to the predicted occupancy situation, a conflict constraint matrix is ​​constructed, and each element in the matrix represents the occupancy conflict situation of the corresponding vehicle in the corresponding time slice and virtual lane;

[0026] Using a dynamic programming algorithm, we traverse each time slice in the time dimension and determine the occupancy of each vehicle in the virtual lane dimension.

[0027] If there are multiple vehicles occupying conflicts in the current time slice and virtual lane, the paths of the conflicting vehicles are locally adjusted according to the conflict constraint matrix;

[0028] Through iterative optimization, the occupancy conflicts between vehicles are continuously eliminated until there are no conflicts in all time slices;

[0029] Finally, a conflict-free multi-vehicle path planning solution that satisfies the conflict constraint matrix is ​​obtained.

[0030] According to the logistics park vehicle path planning method provided by the present invention, high-precision map data of the park is obtained, road network topology, lane lines, and traffic sign marking elements are extracted, lane-level refinement of the optimal path is performed, and a driving trajectory is generated, including:

[0031] Obtain high-precision map data pre-established in the park, and extract key elements of the road network topology, lane alignment, traffic signs and markings in the map through image processing algorithms;

[0032] According to the obtained road network topology, the graph theory algorithm is used to determine the optimal path between the starting point and the end point;

[0033] For the determined optimal path, the path is refined at the lane level by analyzing the extracted lane linear data to obtain a detailed lane-level driving route;

[0034] According to the detailed lane-level driving route, combined with the traffic signs and markings position information, a trajectory interpolation algorithm is used to generate a smooth driving trajectory;

[0035] Perform collision detection on the generated driving trajectory to determine whether the trajectory collides with obstacles. If there is a collision, return to replan the path until there is no collision;

[0036] According to the dynamic characteristics of the vehicle, the speed of the smooth trajectory is planned to obtain a speed curve that meets the vehicle dynamic constraints;

[0037] The planned driving trajectory and speed curve are used as control instructions for autonomous driving and passed to the vehicle's underlying control system for execution.

[0038] According to the logistics park vehicle path planning method provided by the present invention, the high-precision position and posture information of the vehicle is obtained through multi-sensor fusion positioning technology, which is matched with the driving trajectory to calculate the trajectory tracking error, including:

[0039] The high-precision position and attitude information of the vehicle is obtained through multi-sensor fusion positioning technology. The sensors include but are not limited to GPS, IMU, wheel speed meter and laser radar. The Kalman filter algorithm is used to fuse the data of each sensor to obtain the precise position and real-time attitude of the vehicle;

[0040] Generate a reference driving trajectory based on high-precision map data and the driving status of the vehicle, where the reference driving trajectory is an ideal driving path of the vehicle on the road, including position, speed, acceleration and curvature information;

[0041] The vehicle position and posture information obtained by multi-sensor fusion is matched with the reference driving trajectory. The nearest point matching algorithm is used to find the projection point of the vehicle's current position on the reference trajectory, and the trajectory tracking error between the vehicle position and the projection point is calculated.

[0042] According to the logistics park vehicle path planning method provided by the present invention, the vehicle exchanges information with other vehicles and roadside facilities through V2X communication technology, perceives the surrounding environment in real time, and activates the collision warning and obstacle avoidance mechanism if other vehicles or obstacles are detected ahead, including:

[0043] The vehicle is equipped with a V2X communication module to communicate wirelessly with other vehicles and roadside facilities to exchange vehicle location and speed information;

[0044] Vehicles use lidar and millimeter-wave radar to collect real-time data about the surrounding environment and perceive road conditions;

[0045] Based on the acquired vehicle information and environmental data, a deep learning algorithm is used to identify and classify obstacles ahead to determine whether they are vehicles or other obstacles.

[0046] If a vehicle or obstacle is identified ahead, the distance and speed difference parameters between the vehicle and the obstacle are calculated;

[0047] The collision risk is determined by the decision algorithm. If the distance is less than the safety threshold and the speed difference is greater than the set value, it is determined that there is a collision risk;

[0048] When a collision risk is determined, the control system sends out a collision warning signal and plans an obstacle avoidance path at the same time;

[0049] Control the steering and braking actuators to make the vehicle drive along the planned obstacle avoidance path, avoid obstacles and ensure driving safety.

[0050] On the other hand, to achieve the above-mentioned purpose, the present invention also provides a logistics park vehicle path planning system, including: a directed weighted graph generation module, an optimal path calculation module, a conflict constraint matrix construction module, a driving path generation module, a trajectory tracking error calculation module and a warning obstacle avoidance module;

[0051] The directed weighted graph generation module is used to obtain the real-time location and task status information of vehicles in the park, dynamically build a vehicle path planning network model, divide the road into nodes and edges, and form a directed weighted graph;

[0052] The optimal path calculation module is used to calculate the optimal path for each vehicle based on the directed weighted graph using an improved multi-lane algorithm and taking into account the path length, congestion level and vehicle characteristics;

[0053] The conflict constraint matrix building module is used to introduce virtual lanes and time dimensions for the optimal path, predict the occupancy of multiple vehicles on the virtual lanes for each time slice, and build a conflict constraint matrix;

[0054] The driving path generation module is used to obtain high-precision map data of the park, extract road network topology, lane lines, and traffic sign marking elements, perform lane-level refinement on the optimal path, and generate a driving trajectory;

[0055] The trajectory tracking error calculation module is used to obtain the high-precision position and posture information of the vehicle through multi-sensor fusion positioning technology, match the high-precision position and posture information of the vehicle with the driving trajectory, and calculate the trajectory tracking error;

[0056] The warning and obstacle avoidance module is used for the vehicle to exchange information with other vehicles and roadside facilities through V2X communication technology, perceive the surrounding environment in real time, and activate the collision warning and obstacle avoidance mechanism if other vehicles or obstacles are detected ahead.

[0057] Technical effect of the invention: The invention discloses a vehicle path planning method and system for a logistics park, which plans the optimal distribution route through a scientific method to ensure that the goods can be distributed from the warehouse to various destinations in the shortest time and at the lowest cost. The driving routes and sequences of vehicles are reasonably planned to reduce empty driving and repeated driving, and improve the loading rate and transportation efficiency of vehicles. Unnecessary transportation mileage and time are reduced, thereby reducing fuel consumption, labor costs and vehicle wear and tear, and achieving cost savings. Ensure that the goods are delivered on time and safely, improve the punctuality and reliability of distribution, and enhance customer satisfaction with logistics services. Through accurate planning of vehicle paths, logistics park managers can better monitor and dispatch vehicles and improve the operation and management level of the park. During the distribution process, various emergencies may be encountered, such as traffic congestion, weather changes, etc., and path planning needs to be able to respond quickly to these changes and re-plan routes. Reasonably allocate vehicle and driver resources to avoid resource waste and ensure that vehicles and personnel in the logistics park are fully utilized. Plan safe driving routes, abide by traffic rules, reduce the risk of traffic accidents, and ensure the safety of personnel and goods. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0059] Figure 1 A schematic diagram of a process flow of a logistics park vehicle path planning method according to an embodiment of the present invention;

[0060] Figure 2 A schematic diagram of a process for forming a directed weighted graph according to an embodiment of the present invention;

[0061] Figure 3 A schematic diagram of a process for calculating the optimal path for each vehicle according to an embodiment of the present invention;

[0062] Figure 4 A schematic diagram of a process for constructing a conflict constraint matrix according to an embodiment of the present invention;

[0063] Figure 5 A schematic diagram of a process for generating a driving trajectory according to an embodiment of the present invention;

[0064] Figure 6 A schematic diagram of a flow chart of calculating a trajectory tracking error according to an embodiment of the present invention;

[0065] Figure 7A schematic diagram of a process in which a vehicle in an embodiment of the present invention exchanges information with other vehicles and roadside facilities through V2X communication technology to perceive the surrounding environment in real time;

[0066] Figure 8 The present invention is a schematic diagram of the structure of a vehicle path planning system for a logistics park according to an embodiment of the present invention. DETAILED DESCRIPTION

[0067] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0068] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0069] like Figure 1 As shown, this embodiment provides a logistics park vehicle path planning method, including:

[0070] Obtain the real-time location and mission status information of vehicles in the park, dynamically build a vehicle path planning network model, divide the road into nodes and edges, and form a directed weighted graph;

[0071] Based on the directed weighted graph, an improved multi-lane algorithm is used to calculate the optimal path for each vehicle by comprehensively considering the path length, congestion level and vehicle characteristics;

[0072] For the optimal path, virtual lanes and time dimensions are introduced to predict the occupancy of multiple vehicles on the virtual lanes for each time slice and build a conflict constraint matrix.

[0073] Obtain high-precision map data of the park, extract road network topology, lane lines, traffic signs and markings, perform lane-level refinement on the optimal path, and generate driving trajectories;

[0074] Through multi-sensor fusion positioning technology, the high-precision position and posture information of the vehicle is obtained, the high-precision position and posture information of the vehicle is matched with the driving trajectory, and the trajectory tracking error is calculated;

[0075] Vehicles exchange information with other vehicles and roadside facilities through V2X communication technology, and perceive the surrounding environment in real time. If other vehicles or obstacles are detected ahead, collision warning and obstacle avoidance mechanisms are activated.

[0076] Further, such as Figure 2As shown in the figure, the real-time location and task status information of vehicles in the park are obtained, and a vehicle path planning network model is dynamically constructed. The roads are divided into nodes and edges to form a directed weighted graph, including:

[0077] The vehicle positioning device is used to obtain the location coordinates of each vehicle in the park and the status information of the currently executed task in real time, and upload them to the path planning server;

[0078] According to the obtained vehicle position coordinates, the road section where the vehicle is located is marked on the map, the road is divided into several nodes and directed edges connecting the nodes, and a vehicle path planning network model is constructed;

[0079] The constructed network model is analyzed using graph theory algorithms to calculate the weight of each edge. The weight value needs to be set by comprehensively considering factors such as road length and traffic status to form a directed weighted graph for vehicle path planning.

[0080] Specifically, the position coordinates of each vehicle in the park and the status information of the currently executed tasks are obtained in real time through the vehicle positioning device, and uploaded to the path planning server. According to the obtained vehicle position coordinates, the road section where the vehicle is located is marked on the map, and the road is divided into several nodes and directed edges connecting the nodes to construct a vehicle path planning network model. The constructed network model is analyzed by graph theory algorithm to calculate the weight value of each edge. The setting of the weight value needs to comprehensively consider factors such as road length and traffic status, so as to form a directed weighted graph model for vehicle path planning. For the starting position and destination of each vehicle, the shortest path algorithm, such as Dijkstra algorithm or A algorithm, is applied to the constructed directed weighted graph to search for one or more optimal paths from the starting point to the end point. If the vehicle deviates from the established path or encounters traffic congestion during driving, the dynamic update of the path is triggered, and the path is replanned according to the location of the vehicle to guide the vehicle to adjust the driving route in time. By continuously tracking the vehicle's driving trajectory and real-time monitoring of task execution, the weight values ​​of each edge in the network model are dynamically updated, and the path planning model is optimized using an incremental learning algorithm to improve the accuracy and real-time performance of path planning. The vehicle scheduling decision model is constructed by comprehensively utilizing multi-source data such as vehicle location, road information, and task status. While meeting the timeliness requirements for task completion, it also takes into account vehicle energy consumption and road traffic efficiency, achieving global optimization of vehicle scheduling and path planning.

[0081] Further, such as Figure 3 As shown in the figure, based on the directed weighted graph, an improved multi-lane algorithm is used to comprehensively consider the path length, congestion level, and vehicle characteristics to calculate the optimal path for each vehicle, including:

[0082] According to the structure and weight information of the directed weighted graph, a road network topology model is constructed and stored in the graph database;

[0083] Obtain real-time traffic flow data, dynamically evaluate the degree of road congestion, and associate the congestion index with the road section for storage;

[0084] For the starting point and end point of each vehicle, the improved Dijkstra algorithm is used to calculate multiple alternative routes based on the path length and congestion index;

[0085] According to the characteristic parameters of the vehicle, the adaptability of the alternative routes is scored, and the routes with high matching degree with the vehicle characteristics are selected;

[0086] The alternative paths are further optimized through the ant colony optimization algorithm, which simulates the vehicle interaction behavior in multi-lane scenarios and obtains more reasonable path planning results.

[0087] Match the optimized path with the vehicle information to generate the optimal driving route for each vehicle, and return the result to the vehicle terminal device;

[0088] Continuously monitor changes in road conditions and vehicle positions. If congestion worsens or the vehicle deviates from the planned route, it triggers path replanning to ensure real-time dynamic optimization.

[0089] Specifically, firstly, a road network topology model is constructed according to the structure and weight information of the directed weighted graph. For example, using the Neo4j graph database, the road sections are represented as nodes, and the connection relationship between the road sections is represented as edges. The edges are attached with weight information such as the length of the road section, speed limit, number of lanes, etc. Then, real-time traffic flow data is obtained, and the vehicle trajectory data and road monitoring video are used to calculate the traffic density of each road section through density estimation algorithms such as kernel density estimation (KDE). Then, combined with the number of lanes and length of the road section, the congestion index is obtained. For example, congestion index = traffic density × road section length ÷ number of lanes, and the congestion index ranges from 0 to 10. Then, for the starting point and end point of each vehicle, the improved Dijkstra algorithm is used. On the basis of the shortest path, the congestion index is used as an additional weight to calculate the top five alternative paths that comprehensively consider the path length and smoothness. Then, according to the characteristic parameters of the vehicle, such as vehicle size, volume, load capacity, etc., the adaptability of the alternative paths is scored. For example, large trucks are suitable for highways, small passenger cars are suitable for branch roads, etc., and the three paths with the highest matching degree are screened out. The path is further optimized through the ant colony optimization algorithm. 50 ants are set and iterated 100 times to simulate the vehicle interaction behaviors in multi-lane scenarios, such as lane changing and overtaking, to obtain more reasonable path planning results. Finally, the optimized path is matched with the vehicle information to generate the optimal driving route for each vehicle. For example, truck routes try to avoid congested sections and height-restricted sections, and passenger routes try to choose routes with good scenery, etc., and the results are returned to the on-board terminal device in the form of navigation information. The system also needs to continuously monitor road conditions and vehicle position changes. If it is found that the congestion index of a certain section exceeds 8 or the vehicle deviates from the established route by more than 10 kilometers, the path re-planning is triggered. The latest road condition information and vehicle position are used to dynamically optimize the path in real time to ensure driving safety and efficiency.

[0090] Further, such as Figure 4 As shown in the figure, for the optimal path, virtual lanes and time dimensions are introduced, the occupancy of multiple vehicles on the virtual lanes is predicted for each time slice, and a conflict constraint matrix is ​​constructed, including:

[0091] According to the optimal path, obtain the virtual lane information and time dimension information on the path;

[0092] For each time slice, obtain the predicted occupancy of multiple vehicles on the virtual lane;

[0093] According to the predicted occupancy situation, a conflict constraint matrix is ​​constructed, and each element in the matrix represents the occupancy conflict situation of the corresponding vehicle in the corresponding time slice and virtual lane;

[0094] Using a dynamic programming algorithm, we traverse each time slice in the time dimension and determine the occupancy of each vehicle in the virtual lane dimension.

[0095] If there are multiple vehicles occupying conflicts in the current time slice and virtual lane, the paths of the conflicting vehicles are locally adjusted according to the conflict constraint matrix;

[0096] Through iterative optimization, the occupancy conflicts between vehicles are continuously eliminated until there are no conflicts in all time slices;

[0097] Finally, a conflict-free multi-vehicle path planning solution that satisfies the conflict constraint matrix is ​​obtained.

[0098] Specifically, virtual lane information and time dimension information are obtained based on the optimal path. For example, an optimal path passes through virtual lanes numbered 1, 2, and 3, with a time span of 10 time slices. For these 10 time slices, the occupancy of multiple vehicles on virtual lanes 1, 2, and 3 is predicted. For example, in the first time slice, it is predicted that two vehicles A and B will occupy lane 1, and vehicle C will occupy lane 2, while no vehicle will occupy lane 3. Based on this, a 10*3 conflict constraint matrix is ​​constructed. The value of the matrix element (1,1) is 2, indicating that there is an occupation conflict between two vehicles in lane 1 in the first time slice, the value of (1,2) is 0, indicating no conflict, and the value of (1,3) is 0, and so on. Then, dynamic programming is used to traverse from the first time slice, and the occupation conflict of lanes 1, 2, and 3 is judged in each time slice. For example, in the first time slice, if the value of (1,1) is 2, there is an occupation conflict between cars A and B. Then, according to the conflict constraint matrix, the path of A or B is partially adjusted, such as adjusting car A to lane 2 to eliminate the occupation conflict, updating the conflict constraint matrix, and then judging the next time slice. By iteratively optimizing 10 time slices, a conflict-free multi-vehicle path planning scheme that satisfies the conflict constraint matrix is ​​finally obtained. This scheme can ensure that when cars A, B, and C pass through lanes 1, 2, and 3 within 10 time slices, there will be no occupation conflict in any time slice, thereby achieving safe and efficient multi-vehicle collaboration.

[0099] Further, such as Figure 5 As shown in the figure, obtain high-precision map data of the park, extract road network topology, lane lines, and traffic sign markings, perform lane-level refinement on the optimal path, and generate a driving trajectory, including:

[0100] Obtain high-precision map data pre-established in the park, and extract key elements of the road network topology, lane alignment, traffic signs and markings in the map through image processing algorithms;

[0101] According to the obtained road network topology, the graph theory algorithm is used to determine the optimal path between the starting point and the end point;

[0102] For the determined optimal path, the path is refined at the lane level by analyzing the extracted lane linear data to obtain a detailed lane-level driving route;

[0103] According to the detailed lane-level driving route, combined with the traffic signs and markings position information, a trajectory interpolation algorithm is used to generate a smooth driving trajectory;

[0104] Perform collision detection on the generated driving trajectory to determine whether the trajectory collides with obstacles. If there is a collision, return to replan the path until there is no collision;

[0105] According to the dynamic characteristics of the vehicle, the speed of the smooth trajectory is planned to obtain a speed curve that meets the vehicle dynamic constraints;

[0106] The planned driving trajectory and speed curve are used as control instructions for autonomous driving and passed to the vehicle's underlying control system for execution.

[0107] Specifically, first, a high-precision laser radar scanner is used to obtain three-dimensional point cloud data in the park, and the point cloud density reaches more than 100 points per square meter. Then, the image semantic segmentation algorithm based on deep learning is used to process the point cloud, extract key elements such as roads, lane lines, and traffic signs, and construct a semantic map. Then, the A* search algorithm is used to search for the optimal path between the starting point and the end point on the semantic map, and a global optimal path is obtained by comprehensively considering factors such as path length, travel time, and traffic flow. On this basis, the global path is refined using the lane line extraction results to generate a locally optimal lane-level path planning, and a smooth driving trajectory is fitted using the cubic spline interpolation algorithm. To ensure driving safety, the trajectory is collided, and the collision risk between the trajectory and surrounding obstacles is detected using the layered bounding box method. Once a collision hazard is found, it is returned for replanning. Finally, the trajectory is speed-planned based on the vehicle dynamics model. By solving the unconstrained optimization problem, a smooth and efficient speed curve is obtained under the condition of satisfying the lateral and longitudinal acceleration constraints. The planning results are converted into control instructions and sent to the bottom layer of the vehicle to achieve precise automatic driving.

[0108] Further, such as Figure 6 As shown in the figure, through multi-sensor fusion positioning technology, the high-precision position and posture information of the vehicle is obtained, matched with the driving trajectory, and the trajectory tracking error is calculated, including:

[0109] The high-precision position and attitude information of the vehicle is obtained through multi-sensor fusion positioning technology. The sensors include but are not limited to GPS, IMU, wheel speed meter and laser radar. The Kalman filter algorithm is used to fuse the data of each sensor to obtain the precise position and real-time attitude of the vehicle;

[0110] Generate a reference driving trajectory based on high-precision map data and the vehicle's driving status. This trajectory is the ideal driving path of the vehicle on the road, including position, speed, acceleration and curvature information;

[0111] The vehicle position and posture information obtained by multi-sensor fusion is matched with the reference driving trajectory. The nearest point matching algorithm is used to find the projection point of the vehicle's current position on the reference trajectory, and the trajectory tracking error between the vehicle position and the projection point is calculated.

[0112] Specifically, the high-precision position and attitude information of the vehicle is obtained through multi-sensor fusion positioning technology, among which the positioning accuracy of GPS can reach within 1 meter, the attitude measurement accuracy of IMU can reach 1 degree, the speed measurement accuracy of wheel speed meter can reach 1 meter / second, and the obstacle detection range of laser radar can reach 100 meters. The Kalman filter algorithm is used to fuse these sensor data, and the vehicle kinematic model and observation model are established to recursively estimate the position, speed and attitude of the vehicle. The positioning accuracy after fusion can reach 1 meter. According to the high-precision map data and the driving status of the vehicle, the cubic spline curve fitting algorithm is used to generate the reference driving trajectory. The position accuracy of the trajectory can reach 0.5 meters, which contains the ideal position, speed, acceleration and turning radius of the vehicle on the road. The real-time vehicle position and attitude obtained by multi-sensor fusion are matched with the reference trajectory. The nearest point matching algorithm is used to search for the point on the reference trajectory with the closest Euclidean distance to the current position of the vehicle as the matching point, and the lateral deviation and longitudinal deviation between the vehicle position and the matching point are calculated. If the lateral deviation exceeds 2 meters or the longitudinal deviation exceeds 1 meter, it is judged that the vehicle has a tracking error and deviates from the reference trajectory. According to the size and direction of the tracking error, the model predictive control algorithm is used to establish a vehicle dynamics model, predict the vehicle's motion trajectory in the next 2 seconds, and optimize the calculation of the vehicle's steering angle and acceleration control to minimize the deviation between the predicted trajectory and the reference trajectory. The control cycle is 0.5 seconds. By executing steering and acceleration and deceleration instructions, the vehicle returns to the reference trajectory within 2 seconds, and the tracking error is reduced to less than 1 meter. During the trajectory tracking control process, the vehicle position and posture are continuously obtained with a cycle of 0.1 seconds, the tracking error is calculated in real time, and the steering angle and acceleration are adjusted according to the error size until the tracking error converges to less than 0.5 meters. If the vehicle cannot return to the reference trajectory for 5 consecutive seconds and the tracking error continues to be greater than 5 meters, it is judged that there may be a sensor failure or positioning failure, and the exception handling mechanism is activated to reduce the speed and try to reposition. If it still cannot recover within 10 seconds, the vehicle is controlled to stop and alarm, and switch to manual driving mode to ensure driving safety.

[0113] Further, such as Figure 7 As shown in the figure, the vehicle exchanges information with other vehicles and roadside facilities through V2X communication technology to perceive the surrounding environment in real time. If other vehicles or obstacles are detected ahead, the collision warning and obstacle avoidance mechanism is activated, including:

[0114] The vehicle is equipped with a V2X communication module to communicate wirelessly with other vehicles and roadside facilities to exchange vehicle location and speed information;

[0115] Vehicles use lidar and millimeter-wave radar to collect real-time data about the surrounding environment and perceive road conditions;

[0116] Based on the acquired vehicle information and environmental data, a deep learning algorithm is used to identify and classify obstacles ahead to determine whether they are vehicles or other obstacles.

[0117] If a vehicle or obstacle is identified ahead, the distance and speed difference parameters between the vehicle and the obstacle are calculated;

[0118] The collision risk is determined by the decision algorithm. If the distance is less than the safety threshold and the speed difference is greater than the set value, it is determined that there is a collision risk;

[0119] When a collision risk is determined, the control system sends out a collision warning signal and plans an obstacle avoidance path at the same time;

[0120] Control the steering and braking actuators to make the vehicle drive along the planned obstacle avoidance path, avoid obstacles and ensure driving safety.

[0121] Specifically, the vehicle is equipped with a V2X communication module, and uses wireless communication technologies such as DSRC or C-V2X to establish a dedicated short-range communication link with surrounding vehicles and roadside facilities, exchanging the vehicle's GPS position, speed, acceleration, heading angle and other information every 100 milliseconds, with a communication distance of up to 300 meters. The vehicle is equipped with a Velodyne 64-line laser radar, which scans the environment within a range of 200 meters at a speed of 1 million points per second to generate a three-dimensional point cloud image; it is also equipped with a Continental ARS408 long-range millimeter-wave radar, which operates in the 76-77GHz frequency band and has a detection distance of up to 250 meters. After the sensor collects environmental data, it transmits it to the central processor via the CAN bus. The central processor uses the NVIDIA Drive PX 2 autonomous driving computing platform to identify and classify obstacles based on a convolutional neural network with an accuracy rate of more than 96%. If an obstacle such as a vehicle or pedestrian is identified 20-30 meters in front of the vehicle, the distance to the obstacle is measured by the laser radar and millimeter-wave radar, and the obstacle speed information is obtained through V2X communication to calculate the speed difference between the two vehicles. Based on the TTC (collision time) collision risk assessment model, assuming that the speed of the vehicle is 50km / h, the speed of the front vehicle is 30km / h, and the distance between the two vehicles is 20 meters, the calculated TTC is (20 meters) / (50km / h-30km / h) = 8 seconds. If the TTC is less than the set safety threshold (such as 5 seconds), it is judged that there is a collision risk. When the collision risk is confirmed, the control system issues a collision warning to the driver through the speaker or HUD head-up display, and at the same time combines the road line shape and the trajectory of surrounding vehicles to plan the obstacle avoidance path through the artificial potential field method. The path planning results are transmitted to the EPS (electronic power steering system) through the steering controller to control the steering wheel angle; transmitted to the ESC (electronic stability control system) through the deceleration controller to independently control the braking force of the four wheels, so that the vehicle can safely avoid obstacles.

[0122] like Figure 8 As shown, this embodiment also provides a logistics park vehicle path planning system, including:

[0123] Directed weighted graph generation module, optimal path calculation module, conflict constraint matrix construction module, driving path generation module, trajectory tracking error calculation module and early warning obstacle avoidance module;

[0124] Directed weighted graph generation module, used to obtain the real-time location and task status information of vehicles in the park, dynamically build a vehicle path planning network model, divide the road into nodes and edges, and form a directed weighted graph;

[0125] The optimal path calculation module is used to calculate the optimal path for each vehicle based on the directed weighted graph, using an improved multi-lane algorithm, taking into account the path length, congestion level and vehicle characteristics;

[0126] Construct a conflict constraint matrix module to introduce virtual lanes and time dimensions for the optimal path, predict the occupancy of multiple vehicles on the virtual lanes for each time slice, and construct a conflict constraint matrix;

[0127] The driving path generation module is used to obtain high-precision map data of the park, extract road network topology, lane lines, traffic signs and markings, perform lane-level refinement on the optimal path, and generate a driving trajectory;

[0128] The trajectory tracking error calculation module is used to obtain the high-precision position and posture information of the vehicle through multi-sensor fusion positioning technology, match the high-precision position and posture information of the vehicle with the driving trajectory, and calculate the trajectory tracking error;

[0129] The early warning and obstacle avoidance module is used for vehicles to exchange information with other vehicles and roadside facilities through V2X communication technology, perceive the surrounding environment in real time, and activate the collision warning and obstacle avoidance mechanism if other vehicles or obstacles are detected ahead.

[0130] The present invention discloses a vehicle path planning method and system for a logistics park. The optimal distribution route is planned by a scientific method to ensure that the goods can be distributed from the warehouse to various destinations in the shortest time and at the lowest cost. The driving route and sequence of the vehicle are reasonably planned to reduce empty driving and repeated driving, and improve the loading rate and transportation efficiency of the vehicle. Unnecessary transportation mileage and time are reduced, thereby reducing fuel consumption, labor costs and vehicle wear and tear, and achieving cost savings. It is ensured that the goods are delivered on time and safely, the punctuality and reliability of distribution are improved, and customer satisfaction with logistics services is enhanced. Through accurate planning of vehicle paths, logistics park managers can better monitor and dispatch vehicles and improve the operation and management level of the park. In the distribution process, various emergencies may be encountered, such as traffic congestion, weather changes, etc., and path planning needs to be able to respond to these changes quickly and re-plan the route. Data support is provided for the operational decision-making of the logistics park, such as vehicle scheduling, warehouse storage strategy, human resource allocation, etc. Energy consumption and carbon emissions are reduced by optimizing routes, and green logistics and sustainable development are supported. Vehicle and driver resources are reasonably allocated to avoid resource waste, and ensure that vehicles and personnel in the logistics park are fully utilized. Plan safe driving routes, abide by traffic rules, reduce the risk of traffic accidents, and ensure the safety of people and goods.

[0131] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for vehicle path planning in a logistics park, characterized in that: include: Obtain the real-time location and mission status information of vehicles in the park, dynamically build a vehicle path planning network model, divide the road into nodes and edges, and form a directed weighted graph; According to the directed weighted graph, an improved multi-lane algorithm is used to comprehensively consider the path length, congestion level and vehicle characteristics to calculate the optimal path for each vehicle; For the optimal path, virtual lanes and time dimensions are introduced, the occupancy of multiple vehicles on the virtual lanes is predicted for each time slice, and a conflict constraint matrix is ​​constructed; Obtain high-precision map data of the park, extract road network topology, lane lines, and traffic sign marking elements, perform lane-level refinement on the optimal path, and generate a driving trajectory; The high-precision position and posture information of the vehicle is obtained through multi-sensor fusion positioning technology, the high-precision position and posture information of the vehicle is matched with the driving trajectory, and the trajectory tracking error is calculated; Vehicles exchange information with other vehicles and roadside facilities through V2X communication technology, and perceive the surrounding environment in real time. If other vehicles or obstacles are detected ahead, collision warning and obstacle avoidance mechanisms are activated.

2. The method for logistics park vehicle path planning according to claim 1, characterized in that: Obtain the real-time location and mission status information of vehicles in the park, dynamically build a vehicle path planning network model, divide the road into nodes and edges, and form a directed weighted graph, including: The vehicle positioning device is used to obtain the location coordinates of each vehicle in the park and the status information of the currently executed task in real time, and upload them to the path planning server; According to the obtained vehicle position coordinates, the road section where the vehicle is located is marked on the map, the road is divided into several nodes and directed edges connecting the nodes, and a vehicle path planning network model is constructed; The constructed network model is analyzed using graph theory algorithms to calculate the weight of each edge. The weight value needs to be set by comprehensively considering factors such as road length and traffic status to form a directed weighted graph for vehicle path planning.

3. The logistics park vehicle path planning method according to claim 1, characterized in that: According to the directed weighted graph, an improved multi-lane algorithm is used to comprehensively consider the path length, congestion level, and vehicle characteristics to calculate the optimal path for each vehicle, including: According to the structure and weight information of the directed weighted graph, a road network topology model is constructed and stored in a graph database; Obtain real-time traffic flow data, dynamically evaluate the degree of road congestion, and associate the congestion index with the road section for storage; For the starting point and end point of each vehicle, the improved Dijkstra algorithm is used to calculate multiple alternative routes based on the path length and congestion index; According to the characteristic parameters of the vehicle, the adaptability of the alternative routes is scored, and the routes with high matching degree with the vehicle characteristics are selected; The alternative paths are further optimized through the ant colony optimization algorithm, simulating the vehicle interaction behavior in multi-lane scenarios to obtain more reasonable path planning results; Match the optimized path with the vehicle information to generate the optimal driving route for each vehicle, and return the result to the vehicle terminal device; Continuously monitor changes in road conditions and vehicle positions. If congestion worsens or the vehicle deviates from the planned route, it triggers path replanning to ensure real-time dynamic optimization.

4. The logistics park vehicle path planning method according to claim 1, characterized in that: For the optimal path, virtual lanes and time dimensions are introduced to predict the occupancy of multiple vehicles on the virtual lanes for each time slice, and a conflict constraint matrix is ​​constructed, including: According to the optimal path, obtaining virtual lane information and time dimension information on the path; For each time slice, obtain the predicted occupancy of multiple vehicles on the virtual lane; According to the predicted occupancy situation, a conflict constraint matrix is ​​constructed, and each element in the matrix represents the occupancy conflict situation of the corresponding vehicle in the corresponding time slice and virtual lane; Using a dynamic programming algorithm, we traverse each time slice in the time dimension and determine the occupancy of each vehicle in the virtual lane dimension. If there are multiple vehicles occupying conflicts in the current time slice and virtual lane, the paths of the conflicting vehicles are locally adjusted according to the conflict constraint matrix; Through iterative optimization, the occupancy conflicts between vehicles are continuously eliminated until there are no conflicts in all time slices; Finally, a conflict-free multi-vehicle path planning solution that satisfies the conflict constraint matrix is ​​obtained.

5. The method for logistics park vehicle path planning according to claim 1, characterized in that: Obtain high-precision map data of the park, extract road network topology, lane lines, and traffic sign markings, perform lane-level refinement on the optimal path, and generate a driving trajectory, including: Obtain high-precision map data pre-established in the park, and extract key elements of the road network topology, lane alignment, traffic signs and markings in the map through image processing algorithms; According to the obtained road network topology, the graph theory algorithm is used to determine the optimal path between the starting point and the end point; For the determined optimal path, the path is refined at the lane level by analyzing the extracted lane linear data to obtain a detailed lane-level driving route; According to the detailed lane-level driving route, combined with the traffic sign and marking position information, a trajectory interpolation algorithm is used to generate a smooth driving trajectory; Perform collision detection on the generated driving trajectory to determine whether the trajectory collides with obstacles. If there is a collision, return to replan the path until there is no collision; According to the dynamic characteristics of the vehicle, the speed of the smooth trajectory is planned to obtain a speed curve that meets the vehicle dynamic constraints; The planned driving trajectory and speed curve are used as control instructions for autonomous driving and passed to the vehicle's underlying control system for execution.

6. The method for logistics park vehicle path planning according to claim 1, characterized in that: Through multi-sensor fusion positioning technology, the high-precision position and posture information of the vehicle is obtained, matched with the driving trajectory, and the trajectory tracking error is calculated, including: The high-precision position and attitude information of the vehicle is obtained through multi-sensor fusion positioning technology. The sensors include but are not limited to GPS, IMU, wheel speed meter and laser radar. The Kalman filter algorithm is used to fuse the data of each sensor to obtain the precise position and real-time attitude of the vehicle; Generate a reference driving trajectory based on high-precision map data and the driving status of the vehicle, where the reference driving trajectory is an ideal driving path of the vehicle on the road, including position, speed, acceleration and curvature information; The vehicle position and posture information obtained by multi-sensor fusion is matched with the reference driving trajectory. The nearest point matching algorithm is used to find the projection point of the vehicle's current position on the reference trajectory, and the trajectory tracking error between the vehicle position and the projection point is calculated.

7. The method for logistics park vehicle path planning according to claim 1, characterized in that: Vehicles exchange information with other vehicles and roadside facilities through V2X communication technology to perceive the surrounding environment in real time. If other vehicles or obstacles are detected ahead, collision warning and obstacle avoidance mechanisms are activated, including: The vehicle is equipped with a V2X communication module to communicate wirelessly with other vehicles and roadside facilities to exchange vehicle location and speed information; Vehicles use lidar and millimeter-wave radar to collect real-time data about the surrounding environment and perceive road conditions; Based on the acquired vehicle information and environmental data, a deep learning algorithm is used to identify and classify obstacles ahead to determine whether they are vehicles or other obstacles. If a vehicle or obstacle is identified ahead, the distance and speed difference parameters between the vehicle and the obstacle are calculated; The collision risk is determined by the decision-making algorithm. If the distance is less than the safety threshold and the speed difference is greater than the set value, it is determined that there is a collision risk; When a collision risk is determined, the control system sends out a collision warning signal and plans an obstacle avoidance path at the same time; Control the steering and braking actuators to make the vehicle drive along the planned obstacle avoidance path, avoid obstacles and ensure driving safety.

8. A system for the logistics park vehicle path planning method according to any one of claims 1 to 7, characterized in that: include: Directed weighted graph generation module, optimal path calculation module, conflict constraint matrix construction module, driving path generation module, trajectory tracking error calculation module and early warning obstacle avoidance module; The directed weighted graph generation module is used to obtain the real-time location and task status information of vehicles in the park, dynamically build a vehicle path planning network model, divide the road into nodes and edges, and form a directed weighted graph; The optimal path calculation module is used to calculate the optimal path for each vehicle based on the directed weighted graph using an improved multi-lane algorithm and taking into account the path length, congestion level and vehicle characteristics; The conflict constraint matrix building module is used to introduce virtual lanes and time dimensions for the optimal path, predict the occupancy of multiple vehicles on the virtual lanes for each time slice, and build a conflict constraint matrix; The driving path generation module is used to obtain high-precision map data of the park, extract road network topology, lane lines, and traffic sign marking elements, perform lane-level refinement on the optimal path, and generate a driving trajectory; The trajectory tracking error calculation module is used to obtain the high-precision position and posture information of the vehicle through multi-sensor fusion positioning technology, match the high-precision position and posture information of the vehicle with the driving trajectory, and calculate the trajectory tracking error; The warning and obstacle avoidance module is used for the vehicle to exchange information with other vehicles and roadside facilities through V2X communication technology, perceive the surrounding environment in real time, and activate the collision warning and obstacle avoidance mechanism if other vehicles or obstacles are detected ahead.

Images