Intelligent storage AGV path planning method and system based on improved algorithm

Abstract

The invention provides an intelligent storage AGV path planning method and system based on an improved algorithm, and belongs to the technical field of path planning. The method comprises the steps that S1, each AGV controls a sensor array to collect environment data based on a movement instruction; s2, each AGV performs data fusion on each environment data through an intelligent agent to construct an environment map; s3, each AGV plans an advancing path from a starting node to a target node based on an environment map through an improved path planning algorithm; s4, each AGV performs displacement based on the advancing path, deviation correction is performed on the displacement position in the displacement process, and movement information of the obstacle is collected through a sensor array to update the advancing path; s5, the intelligent agents of the AGVs interact displacement information to dynamically update the advancing path until the AGVs move to the target node; and S6, each AGV records a path planning log in real time. The AGV path planning method has the advantages that the environmental adaptability and the global optimization capability of AGV path planning are greatly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of path planning, and particularly relates to an intelligent warehouse AGV path planning method and system based on an improved algorithm. BACKGROUND

[0002] With the rapid development of logistics automation and intelligent manufacturing, an automatic guided vehicle (AGV) is increasingly widely applied in industrial scenarios such as warehousing and production. Path planning, as one of the key technologies for AGV operation, aims to plan an optimal or suboptimal collision-free path for the AGV from a starting point to an ending point.

[0003] Currently, common algorithms for AGV path planning include Dijkstra algorithm, A* algorithm, artificial potential field method, ant colony algorithm and genetic algorithm. The A* algorithm introduces a heuristic search strategy on the basis of the Dijkstra algorithm, estimates the cost from a current node to a target node in addition to the actual cost from the starting point to the current node, thereby significantly improving the search efficiency and helping to find an optimal path, and thus is widely applied. However, the A* algorithm has relatively limited adaptability when facing dynamically appearing obstacles or in a complex environment, and may have a lag in adjusting the planned path and easily fall into a local optimal solution and thus cannot obtain a globally optimal path, usually requiring an additional obstacle avoidance mechanism.

[0004] It should be noted that the artificial potential field method, ant colony algorithm and genetic algorithm also have a tendency to converge to a local optimal solution rather than a globally optimal solution when solving such problems, especially when the path environment is highly complex or the obstacles are dense.

[0005] In summary, in an application environment such as intelligent warehousing, the above algorithms are prone to fall into a local optimal solution in a dynamic complex environment, and the planned path is not globally optimal, which leads to reduced AGV travel efficiency, increased transportation time and cost. Therefore, how to provide an intelligent warehouse AGV path planning method and system based on an improved algorithm to improve the environmental adaptability and global optimization capability of AGV path planning has become a technical problem to be solved. SUMMARY

[0006] The technical problem to be solved by the present application is to provide an intelligent warehouse AGV path planning method and system based on an improved algorithm to improve the environmental adaptability and global optimization capability of AGV path planning.

[0007] In a first aspect, the present application provides an intelligent warehouse AGV path planning method based on an improved algorithm, comprising the following steps:

[0008] Step S1, each AGV acquires an inputted movement instruction carrying a target node, controls a sensor array including a laser radar, a camera and an infrared sensor to collect environmental data based on the movement instruction;

[0009] Step S2, each AGV fuses each of the collected environmental data through a pre-trained and deployed intelligent agent to construct an environmental map;

[0010] Step S3, each AGV plans a travel path from a starting node to the target node based on the environmental map through an improved path planning algorithm;

[0011] Step S4, each AGV moves based on the travel path, corrects the displacement position through an odometer and a gyroscope during the displacement process, collects the motion information of obstacles through the sensor array, and dynamically updates the travel path based on the motion information;

[0012] Step S5, the intelligent agents of each AGV interact with each other to update the travel path based on the displacement information until each AGV moves to the target node;

[0013] Step S6, each AGV records a path planning log in real time and stores the path planning log.

[0014] Further, the step S1 is specifically:

[0015] Each AGV acquires an inputted movement instruction carrying a target node through a 5G communication module, controls a sensor array including a laser radar, a camera and an infrared sensor to collect environmental data including environmental point cloud data, environmental image data and obstacle distance data based on the movement instruction.

[0016] Further, the step S2 is specifically:

[0017] Each AGV initializes an environmental map constructed based on a two-dimensional grid through a pre-trained and deployed intelligent agent, analyzes each of the environmental data to obtain obstacle positions, obstacle sizes and obstacle shapes, and labels whether the two-dimensional grid of the initialized environmental map is passable based on the obstacle positions, the obstacle sizes and the obstacle shapes.

[0018] Further, in the step S3, the path planning algorithm is obtained by improving a heuristic function of an A* algorithm, and the formula of the improved heuristic function is:

[0019] h'(n)=h(n)+k(n);

[0020] k(n)=α / d(n);

[0021] Wherein, h'(n) represents the improved heuristic function; h(n) represents the unimproved heuristic function; k(n) represents the obstacle correction factor of the nth node; a represents the adjustment coefficient; d(n) represents the distance from the nth node to the nearest obstacle;

[0022] The planning of the travel path from the starting node to the target node based on the environment map is specifically:

[0023] Based on the labeling of the environment map, candidate nodes for the next step displacement of the starting node are selected, an open list and a closed list are created, the starting node and each candidate node are added to the open list, and the estimated total cost of each candidate node is initialized:

[0024] f(n) = g(n) + h'(n);

[0025] Wherein, f(n) represents the estimated total cost, i.e. the estimated total distance or the estimated total displacement time length from the nth candidate node to the target node; g(n) represents the actual cost from the starting node to the nth candidate node, i.e. the actual distance or the actual displacement time length;

[0026] The candidate node with the minimum estimated total cost is selected from the open list as the current node, it is judged whether the current node is the target node, if yes, the travel path of each node is recorded, and the path planning is completed; if not, the current node is moved from the open list to the closed list, and the adjacent candidate nodes are expanded and searched, until the current node is the target node, or the open list is empty, and the travel path is recorded.

[0027] Further, the step S4 is specifically:

[0028] Each AGV displaces based on the travel path, acquires actual position information through the odometer and the gyroscope during the displacement process, compares the actual position information and the travel path to correct the displacement position, acquires motion information including at least the motion direction and the motion speed of the obstacle through the sensor array, sets the obstacle avoidance area based on the motion information and the preset obstacle avoidance distance, and dynamically updates the travel path based on the obstacle avoidance area and the latest environment map;

[0029] The step S5 is specifically:

[0030] The agents of each AGV interact with each other through the 5G communication module, at least including displacement information such as AGV position, AGV displacement speed, AGV displacement direction, AGV target node and AGV task state, dynamically update the travel path based on the displacement information, until each AGV displaces to the target node.

[0031] In a second aspect, the present application provides an intelligent warehouse AGV path planning system based on an improved algorithm, comprising the following modules:

[0032] An environment data acquisition module is configured to acquire, by each AGV, an input mobile instruction carrying a target node, and control a sensor array including a laser radar, a camera, and an infrared sensor to acquire environment data based on the mobile instruction;

[0033] An environment map construction module is configured to construct an environment map by each AGV through an agent pre-trained and deployed to fuse each of the acquired environment data.

[0034] A travel path planning module is configured to plan, by each AGV, a travel path from a starting node to a target node based on the environment map through an improved path planning algorithm.

[0035] An AGV displacement module is configured to displace, by each AGV, based on the travel path, and correct a displacement position through an odometer and a gyroscope during the displacement process, acquire motion information of an obstacle through the sensor array, and dynamically update the travel path based on the motion information.

[0036] A displacement information interaction module is configured to interact, by the agent of each AGV, displacement information, and dynamically update the travel path based on the displacement information until each AGV is displaced to the target node.

[0037] A path planning log recording module is configured to record, by each AGV, a path planning log in real time, and store the path planning log.

[0038] Further, the environment data acquisition module is specifically configured to:

[0039] Each AGV acquires, through a 5G communication module, an input mobile instruction carrying a target node, controls a sensor array including a laser radar, a camera, and an infrared sensor to acquire environment data including environment point cloud data, environment image data, and obstacle distance data based on the mobile instruction.

[0040] Further, the environment map construction module is specifically configured to:

[0041] Each AGV initializes an environment map constructed based on a two-dimensional grid through an agent pre-trained and deployed, analyzes each of the environment data to obtain an obstacle position, an obstacle size, and an obstacle shape, and labels whether the two-dimensional grid of the initialized environment map is passable based on the obstacle position, the obstacle size, and the obstacle shape.

[0042] Further, in the travel path planning module, the path planning algorithm is obtained by improving a heuristic function of an A* algorithm, and a formula of the improved heuristic function is:

[0043] h'(n) = h(n) + k(n);

[0044] k(n) = a / d(n);

[0045] wherein h'(n) represents the improved heuristic function; h(n) represents the heuristic function before improvement; k(n) represents an obstacle correction factor of the nth node; a represents an adjustment coefficient; and d(n) represents a distance from the nth node to the nearest obstacle;

[0046] The travel path from the start node to the target node based on the environment map is specifically:

[0047] Based on the label of the environment map, a candidate node for next displacement of the start node is selected, an open list and a closed list are created, the start node and each candidate node are added to the open list, and an estimated total cost of each candidate node is initialized:

[0048] f(n) = g(n) + h'(n);

[0049] wherein f(n) represents an estimated total cost, i.e., an estimated total distance or an estimated total displacement time length from the nth candidate node to the target node; and g(n) represents an actual cost from the start node to the nth candidate node, i.e., an actual distance or an actual displacement time length.

[0050] A candidate node with the minimum estimated total cost is selected from the open list as a current node, it is judged whether the current node is the target node, if yes, travel paths of all nodes are recorded, and path planning is completed; if not, the current node is moved from the open list to the closed list, and the adjacent candidate nodes are expanded for search until the current node is the target node or the open list is empty, and the travel path is recorded.

[0051] Further, the AGV displacement module is specifically used for:

[0052] Each AGV performs displacement based on the travel path, actual position information is acquired through a mileage meter and a gyroscope during the displacement process, the actual position information and the travel path are compared to correct the displacement position, motion information of obstacles including at least motion direction and motion speed is acquired through a sensor array, an obstacle avoidance area is set based on the motion information and a preset obstacle avoidance distance, and the travel path is dynamically updated based on the obstacle avoidance area and the latest environment map;

[0053] The displacement information interaction module is specifically used for:

[0054] The agent of each AGV interacts with each other through a 5G communication module, and at least includes displacement information of AGV position, AGV displacement speed, AGV displacement direction, AGV target node and AGV task state, dynamically updates the travel path based on the displacement information, and until each AGV is displaced to the target node.

[0055] The advantages of the present application are:

[0056] 1. Each AGV obtains an input mobile instruction carrying a target node, controls a sensor array including a laser radar, a camera and an infrared sensor to collect environmental data based on the mobile instruction; each AGV performs data fusion on each collected environmental data through a pre-trained and deployed agent to construct an environmental map, plans a travel path from a starting node to a target node based on the environmental map through an improved path planning algorithm, moves based on the travel path, corrects the displacement position through an odometer and a gyroscope during the displacement process, collects the motion information of obstacles through the sensor array to dynamically update the travel path; the agent of each AGV interacts with each other to update the travel path based on the displacement information until each AGV is displaced to the target node, and records and stores the path planning log in real time; that is, the heuristic function of the A* algorithm is improved, the travel path is planned based on the improved path planning algorithm, the influence of obstacles is fully considered through an obstacle correction factor, the travel path is updated according to the motion information of obstacles and the travel path is updated according to the displacement information of other AGVs during the displacement process, and finally the environmental adaptability and global optimization capability of AGV path planning are greatly improved.

[0057] 2. The sensor array integrating the laser radar, the camera and the infrared sensor simultaneously collects environmental point cloud data, environmental image data and obstacle distance data, realizes omnidirectional and multi-modal perception of the warehouse environment, and this fusion design avoids the limitations of a single sensor (such as the camera being susceptible to light and the radar being insensitive to non-structural obstacles), significantly improves the accuracy of environmental modeling, reduces the risk of false detection and missed detection in complex or variable warehouse scenarios (such as dynamic obstacles), provides reliable input for subsequent path planning, thereby reducing the collision probability of AGV and improving the overall system safety.

[0058] 3. By employing pre-trained and deployed intelligent agents (such as machine learning-based models) to automatically fuse environmental data, the location, size, and shape of obstacles are directly labeled on a two-dimensional grid map, and grid accessibility is determined. This replaces traditional manual setting and static maps, enabling real-time adaptive map updates. Map construction is efficient and lightweight, reducing computational resource consumption. Combined with two-dimensional grid labeling, navigable paths can be quickly generated (labeling "impassable" areas directly avoids dangerous paths), improving planning efficiency and making it easy to extend to new environments (such as changes in warehouse layout), enhancing the versatility and practicality of the solution.

[0059] 4. By improving the heuristic function of the classic A* algorithm, an obstacle correction factor is introduced to dynamically optimize the path, forcing the path to stay away from obstacles. The candidate node expansion and open list optimization algorithms are used to make the path calculation safer, avoid the AGV getting close to obstacles, and reduce the risk of collision in narrow passages. At the same time, the algorithm ensures the optimality of the path by estimating the total cost, which significantly improves the efficiency of warehouse operation.

[0060] 5. During the displacement process, the odometer and gyroscope are used to correct the position (comparing the actual position with the planned path), and obstacle movement information is collected in real time through the sensor array. The obstacle avoidance area is dynamically set and the path is updated. It can adaptively handle sudden changes in the warehouse environment (such as moving people or forklifts) to prevent the AGV from deviating from the path or stopping. At the same time, the preset obstacle avoidance distance can prevent close collisions, improve the continuity of AGV operation, and increase warehouse throughput and equipment life.

[0061] 6. The 5G communication module enables the exchange of displacement information between AGVs (including position, displacement speed, direction, target node, and task status), and dynamically updates the path based on this to avoid conflicts. It supports efficient collaboration among multiple AGVs, rather than isolated planning. In large-scale warehousing systems, it can reduce AGV congestion or deadlock (e.g., by avoiding path intersections) and optimize task scheduling. The low latency of 5G ensures real-time information and is suitable for modern warehouses with multiple AGVs operating in parallel, reducing operating costs.

[0062] 7. By recording and storing path planning logs in real time, a traceable data source is formed. This data, which combines execution data (such as displacement paths) and environmental conditions, facilitates later analysis and provides a data foundation for algorithm optimization (such as training agents based on historical data) and system maintenance (diagnosing fault points or bottlenecks), thereby improving the sustainability of the solution.

[0063] 8. The overall perception and fusion of environmental data are realized through a multi-sensor array (laser radar, camera, infrared sensor), and an accurate environmental map is dynamically constructed in combination with a pre-trained intelligent agent, which significantly improves the adaptability to complex warehouse scenarios; an innovative and optimized path planning algorithm (improved heuristic function of A*) is adopted, which prioritizes obstacle avoidance while ensuring the optimality of the path, effectively enhancing the safety and efficiency of AGV operation; its closed-loop control mechanism (real-time correction of odometry and gyroscope), dynamic obstacle response capability (path update based on motion information), and multi-AGV coordination mechanism (collision avoidance through 5G interaction of displacement information) work together to achieve high robustness and self-optimization of the path in dynamic environments; in addition, the recording of path planning logs provides key data support for system iteration and operation; through the synergistic effect of these technical features, the bottlenecks of traditional AGVs in terms of perception accuracy, path safety, dynamic obstacle avoidance, and multi-vehicle coordination are systematically solved, which can significantly improve the efficiency and reliability of warehouse operations and reduce operation and maintenance costs. BRIEF DESCRIPTION OF DRAWINGS

[0064] The application will be further described below with reference to the accompanying drawings and embodiments.

[0065] Fig. 1 is a flowchart of an intelligent warehouse AGV path planning method based on an improved algorithm.

[0066] Fig. 2 is a structural schematic diagram of an intelligent warehouse AGV path planning system based on an improved algorithm. DETAILED DESCRIPTION

[0067] The technical solution in the embodiments of the present application has the following general idea: the heuristic function of the A* algorithm is improved, and the improved path planning algorithm is used to plan the travel path, the influence of obstacles is fully considered through an obstacle correction factor, and the travel path is updated according to the displacement process, the motion information of obstacles, and the displacement information of other AGVs, so as to improve the environmental adaptability and global optimization capability of AGV path planning.

[0068] Please refer to Figs. 1-2 , a preferred embodiment of an intelligent warehouse AGV path planning method based on an improved algorithm, which includes the following steps:

[0069] Step S1, each AGV obtains an input mobile instruction carrying a target node, and controls a sensor array including a laser radar, a camera, and an infrared sensor to collect environmental data based on the mobile instruction;

[0070] Step S2, each AGV fuses the collected environment data through the pre-trained and deployed agent to construct an environment map;

[0071] Step S3, each AGV plans a travel path from a starting node to a target node based on the environment map through an improved path planning algorithm; the improved path planning algorithm avoids falling into local optimization, enabling the AGV to find a more optimal or even globally optimal travel path, thereby reducing transportation costs and improving transportation efficiency;

[0072] Step S4, each AGV moves based on the travel path, and during the movement, the displacement position is corrected through an odometer and a gyroscope, and the motion information of obstacles is collected through a sensor array, and the travel path is dynamically updated based on the motion information;

[0073] Step S5, the agents of each AGV interact with each other to update the travel path based on the displacement information until each AGV moves to the target node; through a dynamic obstacle avoidance strategy, the AGV can perceive environmental changes in real time during travel, quickly re-plan the travel path, and avoid sudden obstacles, ensuring the safe operation and efficient navigation of the AGV in a dynamic environment;

[0074] Step S6, each AGV records and stores the path planning log in real time, which is used for subsequent path optimization and historical data analysis.

[0075] By recording and storing the path planning log in real time, a traceable data source is formed, combining execution data (such as displacement path) and environmental state, which facilitates later analysis, provides a data basis for algorithm optimization (such as training agents based on historical data) and system maintenance (diagnosing fault points or bottlenecks), and improves the sustainability of the scheme.

[0076] In specific implementation, when the path planning of the AGV falls into a performance bottleneck, the relevant data is uploaded to the cloud server through the 5G communication module for online planning, that is, an efficient cloud-edge-end collaborative working mode is established.

[0077] The step S1 is specifically:

[0078] Each AGV obtains the input mobile instruction carrying the target node through a 5G communication module, controls a sensor array including a laser radar, a camera, and an infrared sensor based on the mobile instruction, and collects environment data including environment point cloud data, environment image data, and obstacle distance data. The laser radar performs 360° rotary scanning around the AGV to obtain the environment point cloud data.

[0079] By integrating the sensor array of laser radar, camera and infrared sensor, collecting the environment point cloud data, environment image data and obstacle distance data at the same time, the omnidirectional and multi-modal perception of the warehouse environment is realized, and the fusion design avoids the limitations of single sensor (such as the camera is easily affected by light, and the radar is not sensitive to non-structural obstacles), which significantly improves the accuracy of environment modeling, reduces the risk of false detection and missed detection in complex or variable warehouse scenes (such as dynamic obstacles), provides reliable input for subsequent path planning, reduces the collision probability of AGV and improves the overall system safety.

[0080] The application optimizes the heuristic function, considers the influence of obstacles, makes the AGV avoid the obstacle dense area, plans a shorter and better path, reduces the transportation cost, and improves the logistics efficiency; at the same time, the improved path planning algorithm enhances the global search ability, reduces the risk of falling into local optimum, improves the reliability and accuracy of the travel path planning. Through the integration of dynamic obstacle avoidance strategy, the AGV can monitor the environmental changes in real time, quickly respond to sudden obstacles and re-plan the travel path, ensure the driving safety, support the prediction of the future position of the obstacle, adjust the travel path in advance, reduce the driving risk, and improve the adaptability in dynamic environment. The optimized heuristic function and dynamic obstacle avoidance strategy improve the search efficiency, reduce the algorithm running time, speed up the travel path planning speed, improve the response speed and work efficiency of the AGV, and have strong stability, which can reduce the system failure rate and ensure the stable operation of the automatic logistics system. Based on the improvement of the traditional A* algorithm, the compatibility with the existing AGV system is good, which can be conveniently integrated into the existing system without the need for large-scale modification of hardware and software, reducing the cost and time, and having good expansibility, which can be combined with other technologies to further optimize the AGV system performance. By monitoring the AGV driving state and position information in real time, the driving direction and speed are adjusted in real time according to the deviation to ensure accurate execution of the planned travel path; by continuously monitoring the environmental changes and updating the path planning in real time, the AGV always travels on the optimal path, improving the adaptability and efficiency in dynamic environment. By giving the AGV stronger autonomous planning and decision-making ability, it can automatically respond to environmental changes and obstacles, reduce the need for manual intervention, reduce the influence of manual operation errors, improve the system automation level and operation efficiency.

[0081] The step S2 is specifically:

[0082] Each AGV initializes an environment map constructed based on a two-dimensional grid through a pre-trained and deployed agent, obtains the obstacle position, obstacle size and obstacle shape through data fusion analysis of each environment data, and labels the two-dimensional grid of the initialized environment map based on the obstacle position, obstacle size and obstacle shape to determine whether it is passable. That is, a two-dimensional grid map model is used to divide the area where the AGV is located into grid cells of equal size, and it is determined whether each grid cell is occupied by an obstacle according to the environment data. If yes, the grid cell is marked as impassable; if not, the grid cell is marked as passable, and the size, shape and other information of the obstacle are recorded.

[0083] By using a pre-trained and deployed agent (such as a machine learning-based model) to automatically perform environment data fusion, directly labeling the obstacle position, size and shape on the two-dimensional grid map, and determining the grid passability, the traditional manual setting and static map are replaced, real-time adaptive map updating is realized, the map construction is efficient and lightweight, and the calculation resource occupation is reduced; combined with two-dimensional grid labeling, a navigable path can be quickly generated (labeling "impassable" area directly avoids dangerous path), which improves the planning efficiency and is easy to extend to new environment (such as warehouse layout change), and enhances the generality and practicality of the scheme.

[0084] In the step S3, the path planning algorithm is obtained by improving the heuristic function of the A* algorithm, and the formula of the improved heuristic function is:

[0085] h'(n)=h(n)+k(n);

[0086] k(n)=α / d(n);

[0087] wherein h'(n) represents the improved heuristic function; h(n) represents the improved heuristic function, which adopts Manhattan distance or Euclidean distance; k(n) represents the obstacle correction factor of the nth node; a represents an adjustment coefficient; d(n) represents the distance from the nth node to the nearest obstacle; during the path planning process, when the node approaches the obstacle, h'(n) increases, so that the path planning algorithm tends to select a path away from the obstacle to avoid falling into local optimum;

[0088] The travel path from the starting node to the target node based on the environment map is specifically:

[0089] Based on the labeling of the environment map, the candidate nodes for the next displacement of the starting node are selected, an open list (OpenList) and a closed list (CloseList) are created, the starting node and each candidate node are added to the open list, and the estimated total cost of each candidate node is initialized.

[0090] f(n) = g(n) + h'(n) ;

[0091] wherein f(n) represents an estimated total cost, i.e. an estimated total distance or an estimated total displacement time length from the nth candidate node to the target node; g(n) represents an actual cost from the starting node to the nth candidate node, i.e. an actual distance or an actual displacement time length;

[0092] selecting a candidate node with the minimum estimated total cost from the open list as a current node, judging whether the current node is the target node, if yes, recording the travel path of each node to complete path planning; if no, moving the current node from the open list to the closed list, and performing an extended search on the adjacent candidate nodes until the current node is the target node or the open list is empty, and recording the travel path. For each adjacent candidate node, if it is in the closed list, it is skipped; if it is not in the open list, its g(n), h'(n) and f(n) are calculated and added to the open list; if it is already in the open list, it is checked whether the path to the adjacent candidate node through the current node is better (i.e. whether g(n) is smaller), if yes, its parent node and the corresponding cost value are updated; the process is repeated until the target node is found or the open list is empty.

[0093] By improving the heuristic function of the classic A* algorithm, a dynamic obstacle correction factor is introduced to optimize the path, forcing the path to move away from the obstacle, and a candidate node expansion and open list optimization algorithm is used to execute, making the path calculation safer, avoiding the AGV approaching the obstacle, and reducing the collision risk in narrow channels; at the same time, the algorithm ensures the optimality of the path through the estimated total cost, significantly improving the efficiency of warehouse operation.

[0094] The step S4 is specifically:

[0095] Each AGV controls the motor drive system and steering system to work based on the travel path to displace, acquires actual position information through the odometer and gyroscope during displacement, compares the actual position information and the travel path to correct the displacement position, acquires motion information of the obstacle including at least the motion direction and the motion speed through the sensor array, sets an obstacle avoidance area (such as a circular or elliptical area) based on the motion information and the preset obstacle avoidance distance, dynamically updates the travel path based on the obstacle avoidance area and the latest environment map; in specific implementation, the Kalman filtering algorithm can be used to predict the motion state of the obstacle to obtain its possible position range in the future t time, for example, for a moving forklift, its position change in the next few seconds is predicted according to its current speed and direction, and the obstacle avoidance area is dynamically set.

[0096] During displacement, the position is corrected (compared with the actual position and the planned path) by combining the odometer and the gyroscope, and the motion information of the obstacles is collected in real time through the sensor array, the obstacle avoidance area is dynamically set and the path is updated, which can adaptively handle the sudden changes (such as moving people or forklifts) in the warehouse environment, avoid the AGV deviating from the path or stalling; at the same time, the pre-set obstacle avoidance distance can prevent close-range collision, improve the continuity of AGV operation, and improve the warehouse throughput and equipment life.

[0097] The step S5 is specifically:

[0098] The agent of each AGV interacts with each other through the 5G communication module, and at least includes displacement information of AGV position, AGV displacement speed, AGV displacement direction, AGV target node and AGV task state, and the travel path is dynamically updated based on the displacement information until each AGV is displaced to the target node.

[0099] The displacement information (including position, displacement speed, direction, target node and task state) between AGVs is realized through the 5G communication module, and the path is dynamically updated based thereon to avoid conflicts, supporting efficient cooperation of multiple AGVs rather than isolated planning, which can reduce AGV congestion or deadlock (such as avoiding path intersection points) in a large warehouse system, and optimize task scheduling; the low delay characteristics of 5G ensure the real-time nature of information, which is suitable for modern warehouses with multiple AGVs operating in parallel, and reduces operating costs.

[0100] Through the multi-sensor array (laser radar, camera, infrared sensor), comprehensive perception and fusion of environmental data are realized, and an accurate environmental map is dynamically constructed in combination with the pre-trained agent, which significantly improves the adaptability to complex warehouse scenes; the innovative and optimized path planning algorithm (improved heuristic function of A*, and introduction of obstacle correction factor) is adopted, which can avoid obstacle areas in priority while ensuring the optimality of the path, effectively enhancing the safety and efficiency of AGV operation; the closed-loop control mechanism (real-time correction of odometer and gyroscope), dynamic obstacle response capability (updating of path based on motion information) and multi-AGV cooperation mechanism (avoiding conflicts by interacting displacement information through 5G) together realize the high robustness self-optimization of the path in the dynamic environment; in addition, the record of the path planning log provides key data support for system iteration and operation and maintenance; through the synergistic effect of a series of technical features, the bottlenecks of traditional AGVs in terms of perception accuracy, path safety, dynamic obstacle avoidance and multi-vehicle cooperation are systematically solved, which can significantly improve the efficiency and reliability of warehouse operation and reduce operation and maintenance costs.

[0101] The preferred embodiment of the intelligent warehouse AGV path planning system based on the improved algorithm comprises the following modules:

[0102] An environmental data collection module is configured to acquire a movement instruction of a target node input by each AGV, and control a sensor array including a laser radar, a camera, and an infrared sensor to collect environmental data based on the movement instruction;

[0103] An environmental map construction module is configured to construct an environmental map by fusing each of the collected environmental data by an agent pre-trained and deployed by each AGV;

[0104] A travel path planning module is configured to plan a travel path from a starting node to a target node by each AGV based on the environmental map by using an improved path planning algorithm, so as to avoid falling into a local optimum by using the improved path planning algorithm, and enable the AGV to find a more optimal or even global optimal travel path, thereby reducing transportation cost and improving transportation efficiency;

[0105] An AGV displacement module is configured to displace each AGV based on the travel path, correct a displacement position by using a mileage counter and a gyroscope during the displacement, collect motion information of an obstacle by using the sensor array, and dynamically update the travel path based on the motion information;

[0106] A displacement information interaction module is configured to interact displacement information between agents of each AGV, dynamically update the travel path based on the displacement information, and displace each AGV to the target node; by using a dynamic obstacle avoidance strategy, the AGV can perceive environmental changes in real time, quickly re-plan a travel path, and avoid a sudden obstacle during travel, thereby ensuring safe operation and efficient navigation of the AGV in a dynamic environment;

[0107] A path planning log recording module is configured to record and store a path planning log in real time by each AGV, and use the path planning log for subsequent path optimization and historical data analysis.

[0108] By recording and storing the path planning log in real time, a traceable data source is formed, which combines execution data (such as a displacement path) and an environmental state, facilitates later analysis, provides a data basis for algorithm optimization (such as training an agent based on historical data) and system maintenance (diagnosing a fault point or a bottleneck), and improves sustainability of a scheme.

[0109] In a specific implementation, when path planning of the AGV falls into a performance bottleneck, relevant data is uploaded to a cloud server for online planning by using a 5G communication module, that is, an efficient cloud-edge-end collaborative working mode is established.

[0110] The environmental data collection module is specifically configured to:

[0111] Each AGV obtains the input mobile instruction carrying the target node through the 5G communication module, controls the sensor array including the laser radar, camera and infrared sensor based on the mobile instruction, and collects environmental data including environmental point cloud data, environmental image data and obstacle distance data. The laser radar performs 360° rotary scanning around the AGV to obtain the environmental point cloud data.

[0112] By integrating the sensor array of laser radar, camera and infrared sensor, simultaneously collecting environmental point cloud data, environmental image data and obstacle distance data, all-around and multi-modal perception of the warehouse environment is realized. This fusion design avoids the limitations of single sensor (such as camera being easily affected by light, radar being not sensitive to non-structural obstacles), significantly improves the accuracy of environmental modeling, reduces the risk of false detection and missed detection in complex or variable warehouse scenarios (such as dynamic obstacles), provides reliable input for subsequent path planning, thereby reducing the collision probability of AGV and improving the overall system safety.

[0113] The present application optimizes the heuristic function, considers the influence of obstacles, makes the AGV avoid the obstacle dense area, plans a shorter and better path, reduces the transportation cost and improves the logistics efficiency. At the same time, the improved path planning algorithm enhances the global search ability, reduces the risk of falling into local optimum, improves the reliability and accuracy of the travel path planning. Through the integration of dynamic obstacle avoidance strategy, the AGV can monitor the environmental changes in real time, quickly respond to sudden obstacles and re-plan the travel path, ensure the driving safety, support the prediction of the future position of the obstacle, adjust the travel path in advance, reduce the driving risk and improve the adaptability in dynamic environment. The optimized heuristic function and dynamic obstacle avoidance strategy improve the search efficiency, reduce the algorithm running time, speed up the travel path planning speed, improve the response speed and working efficiency of the AGV, and have strong stability, which can reduce the system failure rate and ensure the stable operation of the automatic logistics system. Based on the improvement of the traditional A* algorithm, the compatibility with the existing AGV system is good, which can be conveniently integrated into the existing system without the need for large-scale modification of hardware and software, reducing cost and time, and having good expansibility, which can be combined with other technologies to further optimize the performance of the AGV system. By monitoring the AGV driving state and position information in real time, the driving direction and speed are adjusted in real time according to the deviation to ensure accurate execution of the planned travel path. By continuously monitoring the environmental changes and updating the path planning in real time, the AGV always travels on the optimal path, improving the adaptability and efficiency in dynamic environment. By giving the AGV stronger autonomous planning and decision-making ability, it can automatically respond to environmental changes and obstacles, reduce the need for manual intervention, reduce the influence of manual operation errors, and improve the system automation level and operation efficiency.

[0114] The environment map construction module is specifically used for:

[0115] Each AGV initializes an environment map constructed based on a two-dimensional grid through a pre-trained and deployed agent, obtains the obstacle position, obstacle size and obstacle shape through data fusion analysis of each environment data, and labels the two-dimensional grid of the initialized environment map based on the obstacle position, obstacle size and obstacle shape to determine whether it is passable. That is, a two-dimensional grid map model is used to divide the area where the AGV is located into grid cells of equal size, and it is determined whether each grid cell is occupied by an obstacle according to the environment data. If yes, the grid cell is marked as impassable; if not, the grid cell is marked as passable, and the size, shape and other information of the obstacle are recorded.

[0116] By using a pre-trained and deployed agent (such as a machine learning-based model) to automatically perform environment data fusion, directly labeling the obstacle position, size and shape on the two-dimensional grid map, and determining the grid passability, the traditional manual setting and static map are replaced, real-time adaptive map updating is realized, the map construction is efficient and lightweight, the computing resource occupation is reduced; combined with two-dimensional grid labeling, a navigable path can be quickly generated (labeling "impassable" area directly avoids dangerous path), the planning efficiency is improved, and it is easy to extend to new environment (such as warehouse layout change), which enhances the generality and practicality of the scheme.

[0117] In the travel path planning module, the path planning algorithm is obtained by improving the heuristic function of the A* algorithm, and the formula of the improved heuristic function is:

[0118] h'(n)=h(n)+k(n);

[0119] k(n)=α / d(n);

[0120] wherein h'(n) represents the improved heuristic function; h(n) represents the improved heuristic function, which adopts Manhattan distance or Euclidean distance; k(n) represents the obstacle correction factor of the nth node; a represents an adjustment coefficient; d(n) represents the distance from the nth node to the nearest obstacle; during the travel path planning process, when the node is close to the obstacle, h'(n) increases, so that the path planning algorithm tends to select a path away from the obstacle to avoid falling into local optimum;

[0121] The travel path planned from the starting node to the target node based on the environment map is specifically:

[0122] Based on the labeling of the environment map, a candidate node for the next displacement of the starting node is selected, an open list (OpenList) and a closed list (CloseList) are created, the starting node and each candidate node are added to the open list, and the estimated total cost of each candidate node is initialized:

[0123] f(n) = g(n) + h'(n) ;

[0124] wherein f(n) represents an estimated total cost, i.e. an estimated total distance or an estimated total displacement time length from the nth candidate node to the target node; g(n) represents an actual cost from the starting node to the nth candidate node, i.e. an actual distance or an actual displacement time length;

[0125] selecting a candidate node with the minimum estimated total cost from the open list as a current node, judging whether the current node is the target node, if yes, recording the travel path of each node to complete path planning; if no, moving the current node from the open list to the closed list, and performing an extended search on the adjacent candidate nodes until the current node is the target node, or the open list is empty, and recording the travel path. For each adjacent candidate node, if it is in the closed list, it is skipped; if it is not in the open list, its g(n), h'(n) and f(n) are calculated and added to the open list; if it is already in the open list, it is checked whether the path to the adjacent candidate node through the current node is better (i.e. whether g(n) is smaller), if yes, its parent node and the corresponding cost value are updated; the process is repeated until the target node is found or the open list is empty.

[0126] By improving the heuristic function of the classic A* algorithm, a dynamic obstacle correction factor is introduced to optimize the path, forcing the path to move away from obstacles, and a candidate node expansion and open list optimization algorithm is used to execute, making the path calculation safer, avoiding AGV approaching obstacles, and reducing the risk of collision in narrow passages; at the same time, the algorithm ensures the optimality of the path through the estimated total cost, significantly improving the efficiency of warehouse operation.

[0127] The AGV displacement module is specifically used for:

[0128] Each AGV controls the motor drive system and steering system to work based on the travel path to displace, and in the displacement process, the actual position information is obtained through the odometer and the gyroscope, the actual position information and the travel path are compared to correct the displacement position, the motion information of the obstacles at least including the motion direction and the motion speed is collected through the sensor array, the obstacle avoidance area (such as a circular or elliptical area) is set based on the motion information and the preset obstacle avoidance distance, and the travel path is dynamically updated based on the obstacle avoidance area and the latest environment map; in specific implementation, the Kalman filtering algorithm can be used to predict the motion state of the obstacles to obtain their possible position range in the future t time, for example, for a moving forklift, its position change in the next few seconds is predicted according to its current speed and direction, and the obstacle avoidance area is dynamically set.

[0129] The displacement process is combined with the odometer and gyroscope to correct the position (compare the actual position with the planned path), and the sensor array is used to collect real-time obstacle motion information, dynamically set the obstacle avoidance area and update the path, which can adaptively handle sudden changes in the warehouse environment (such as moving people or forklifts), and avoid AGV deviating from the path or stalling; at the same time, the pre-set obstacle avoidance distance can prevent close-range collisions, improve the continuity of AGV operation, and improve the warehouse throughput and equipment life.

[0130] The displacement information interaction module is specifically used for:

[0131] The agent of each AGV interacts with each other through the 5G communication module, and at least includes the displacement information of AGV position, AGV displacement speed, AGV displacement direction, AGV target node and AGV task state, and the travel path is dynamically updated based on the displacement information until each AGV is displaced to the target node.

[0132] The displacement information interaction (including position, displacement speed, direction, target node and task state) between AGVs is realized through the 5G communication module, and the path is dynamically updated based on this to avoid conflicts, supporting efficient collaboration of multiple AGVs rather than isolated planning, which can reduce AGV congestion or deadlock (such as avoiding path intersection points) in large warehouse systems, and optimize task scheduling; the low delay characteristics of 5G ensure the real-time nature of information, which is suitable for modern warehouses with multiple AGVs operating in parallel, and reduces operating costs.

[0133] The comprehensive perception and fusion of environmental data are realized through a multi-sensor array (laser radar, camera, infrared sensor), and an accurate environmental map is dynamically constructed in combination with a pre-trained agent, which significantly improves the adaptability to complex warehouse scenarios; an innovative and optimized path planning algorithm (improved heuristic function of A*, and introduction of obstacle correction factor) is used to ensure the optimality of the path while preferentially avoiding obstacle areas, effectively enhancing the safety and efficiency of AGV operation; the closed-loop control mechanism (real-time correction of odometer and gyroscope), dynamic obstacle response capability (updating of path based on motion information) and multi-AGV collaboration mechanism (avoiding conflicts by interacting displacement information through 5G) together realize the high robustness self-optimization of the path in a dynamic environment; in addition, the record of path planning logs provides key data support for system iteration and operation and maintenance; through the synergistic effect of a series of technical features, the bottlenecks of traditional AGVs in terms of perception accuracy, path safety, dynamic obstacle avoidance and multi-vehicle collaboration are systematically solved, which can significantly improve the efficiency and reliability of warehouse operations and reduce operation and maintenance costs.

[0134] In summary, the advantages of the present application are:

[0135] 1、Through each AGV, the input mobile instruction carrying the target node is obtained, and based on the mobile instruction, the sensor array including a laser radar, a camera and an infrared sensor collects environmental data; each AGV performs data fusion on the collected environmental data through the pre-trained and deployed agent to construct an environmental map, plans a travel path from the starting node to the target node based on the environmental map through an improved path planning algorithm, moves based on the travel path, and in the moving process, the moving position is corrected through an odometer and a gyroscope, and the motion information of the obstacle is collected through the sensor array to dynamically update the travel path; the agents of each AGV interact with each other to update the travel path based on the moving information until each AGV moves to the target node, and the path planning log is recorded and stored in real time; that is, the heuristic function of the A* algorithm is improved, the travel path is planned based on the improved path planning algorithm, the influence of the obstacle is fully considered through an obstacle correction factor, the travel path is updated according to the motion information of the obstacle and the travel path is updated according to the moving information of other AGVs in the moving process, and finally the environmental adaptability and global optimization capability of AGV path planning are greatly improved.

[0136] 2、Through the integration of the sensor array of the laser radar, the camera and the infrared sensor, the environmental point cloud data, the environmental image data and the obstacle distance data are collected at the same time, the omnidirectional and multi-modal perception of the warehouse environment is realized, this fusion design avoids the limitations of a single sensor (such as the camera is easily affected by light, and the radar is not sensitive to non-structural obstacles), significantly improves the accuracy of environmental modeling, reduces the risk of false detection and missed detection in complex or variable warehouse scenes (such as dynamic obstacles), provides reliable input for subsequent path planning, thereby reducing the collision probability of AGV and improving the overall system safety.

[0137] 3、Through the use of pre-trained and deployed agents (such as machine learning-based models) to automatically perform environmental data fusion, the positions, sizes and shapes of obstacles are directly labeled on a two-dimensional grid map, and the grid passability is determined, replacing the traditional manual setting and static map, realizing real-time adaptive map updating, efficient and lightweight map construction, reducing the occupation of computing resources; combined with two-dimensional grid labeling, a navigable path can be quickly generated (labeling “non-passable” areas directly avoids dangerous paths), improving planning efficiency, and easily extended to new environments (such as warehouse layout changes), enhancing the generality and practicality of the scheme.

[0138] 4. By improving the heuristic function of the classic A* algorithm, an obstacle correction factor is introduced to dynamically optimize the path, forcing the path to stay away from obstacles. The candidate node expansion and open list optimization algorithms are used to make the path calculation safer, avoid the AGV getting close to obstacles, and reduce the risk of collision in narrow passages. At the same time, the algorithm ensures the optimality of the path by estimating the total cost, which significantly improves the efficiency of warehouse operation.

[0139] 5. During the displacement process, the odometer and gyroscope are used to correct the position (comparing the actual position with the planned path), and obstacle movement information is collected in real time through the sensor array. The obstacle avoidance area is dynamically set and the path is updated. It can adaptively handle sudden changes in the warehouse environment (such as moving people or forklifts) to prevent the AGV from deviating from the path or stopping. At the same time, the preset obstacle avoidance distance can prevent close collisions, improve the continuity of AGV operation, and increase warehouse throughput and equipment life.

[0140] 6. The 5G communication module enables the exchange of displacement information between AGVs (including position, displacement speed, direction, target node, and task status), and dynamically updates the path based on this to avoid conflicts. It supports efficient collaboration among multiple AGVs, rather than isolated planning. In large-scale warehousing systems, it can reduce AGV congestion or deadlock (e.g., by avoiding path intersections) and optimize task scheduling. The low latency of 5G ensures real-time information and is suitable for modern warehouses with multiple AGVs operating in parallel, reducing operating costs.

[0141] 7. By recording and storing path planning logs in real time, a traceable data source is formed. This data, which combines execution data (such as displacement paths) and environmental conditions, facilitates later analysis and provides a data foundation for algorithm optimization (such as training agents based on historical data) and system maintenance (diagnosing fault points or bottlenecks), thereby improving the sustainability of the solution.

[0142] 8. Through the comprehensive perception and fusion of environmental data by a multi-sensor array (laser radar, camera, infrared sensor), combined with the dynamic construction of an accurate environmental map by a pre-trained agent, the adaptability to complex warehouse scenarios is significantly improved; an innovative and optimized path planning algorithm (improved heuristic function of A*, introduction of obstacle correction factor) is adopted to ensure the optimality of the path while preferentially avoiding obstacle regions, effectively enhancing the safety and efficiency of AGV operation; its closed-loop control mechanism (real-time correction of odometry and gyroscope), dynamic obstacle response capability (path update based on motion information) and multi-AGV coordination mechanism (collision avoidance through 5G interaction of displacement information) together realize the high robustness self-optimization of the path in a dynamic environment; in addition, the record of path planning logs provides key data support for system iteration and operation; through the synergistic effect of a series of technical features, the bottlenecks of traditional AGVs in terms of perception accuracy, path safety, dynamic obstacle avoidance and multi-vehicle coordination are systematically solved, which can significantly improve the efficiency and reliability of warehouse operations and reduce operation and maintenance costs.

[0143] Although the specific embodiments of the present application are described above, those skilled in the art should understand that the specific examples described are only illustrative, and are not intended to limit the scope of the present application, and equivalent modifications and variations made by those skilled in the art in accordance with the spirit of the present application should be covered within the scope of the claims of the present application.

Claims

1. A path planning method for intelligent warehousing AGVs based on an improved algorithm, characterized in that: Includes the following steps: Step S1: Each AGV obtains the input movement command carrying the target node, and controls the sensor array including lidar, camera and infrared sensor to collect environmental data based on the movement command; Step S2: Each AGV uses a pre-trained and deployed agent to fuse the collected environmental data to construct an environmental map; Step S3: Each AGV plans its movement path from the starting node to the target node based on the environmental map using an improved path planning algorithm. Step S4: Each AGV moves according to the travel path. During the movement, the displacement position is corrected by the odometer and gyroscope. The motion information of obstacles is collected by the sensor array, and the travel path is dynamically updated based on the motion information. Step S5: The intelligent agents of each AGV interact with each other to exchange displacement information, and dynamically update the travel path based on the displacement information until each AGV moves to the target node; Step S6: Each AGV records the path planning log in real time and stores the path planning log.

2. The intelligent warehousing AGV path planning method based on an improved algorithm as described in claim 1, characterized in that: Step S1 specifically involves: Each AGV obtains the input movement command carrying the target node through the 5G communication module, and controls the sensor array including lidar, camera and infrared sensor based on the movement command to collect environmental data including environmental point cloud data, environmental image data and obstacle distance data.

3. The intelligent warehousing AGV path planning method based on an improved algorithm as described in claim 1, characterized in that: Step S2 specifically involves: Each AGV initializes an environment map based on a two-dimensional grid through a pre-trained and deployed agent. It then performs data fusion analysis on the environmental data to obtain the obstacle positions, sizes, and shapes. Based on the obstacle positions, sizes, and shapes, it marks the two-dimensional grid of the initialized environment map to indicate whether the obstacle is passable.

4. The intelligent warehousing AGV path planning method based on an improved algorithm as described in claim 1, characterized in that: In step S3, the path planning algorithm is obtained by improving the heuristic function of the A* algorithm. The formula of the improved heuristic function is: h'(n) = h(n) + k(n); k(n) = α / d(n); Where h'(n) represents the improved heuristic function; h(n) represents the original heuristic function; k(n) represents the obstacle correction factor of the nth node; α represents the adjustment coefficient; and d(n) represents the distance from the nth node to the nearest obstacle. The specific steps of planning the movement path from the starting node to the target node based on the environmental map are as follows: Based on the annotations on the environmental map, candidate nodes for the next displacement of the starting node are selected. An open list and a closed list are created. The starting node and each candidate node are added to the open list, and the estimated total cost of each candidate node is initialized. f(n) = g(n) + h'(n); Where f(n) represents the estimated total cost, which is the estimated total distance or estimated total displacement time from the nth candidate node to the target node; g(n) represents the actual cost from the starting node to the nth candidate node, which is the actual distance or actual displacement time. The candidate node with the lowest estimated total cost is selected from the open list as the current node. It is then determined whether the current node is the target node. If it is, the travel path of each node is recorded to complete path planning. If not, the current node is moved from the open list to the closed list, and the adjacent candidate nodes are expanded for search until the current node is the target node or the open list is empty. The travel path is then recorded.

5. The intelligent warehousing AGV path planning method based on an improved algorithm as described in claim 1, characterized in that: Step S4 specifically involves: Each AGV moves based on the travel path. During the movement, it acquires the actual position information through the odometer and gyroscope. It compares the actual position information with the travel path to correct the position. It collects motion information of obstacles, including at least the direction and speed of movement, through the sensor array. It sets the obstacle avoidance area based on the motion information and the preset obstacle avoidance distance. It dynamically updates the travel path based on the obstacle avoidance area and the latest environmental map. Step S5 specifically involves: Each AGV's intelligent agent interacts with each other through a 5G communication module, including at least the AGV's position, AGV's displacement speed, AGV's displacement direction, AGV's target node, and AGV's task status displacement information. Based on the displacement information, the travel path is dynamically updated until each AGV moves to the target node.

6. A smart warehouse AGV path planning system based on an improved algorithm, characterized in that: Includes the following modules: The environmental data acquisition module is used by each AGV to acquire the movement instructions of the target node, and control a sensor array including lidar, camera and infrared sensor to collect environmental data based on the movement instructions. An environment map construction module is used by each AGV to perform data fusion on the collected environmental data through pre-trained and deployed intelligent agents to construct an environment map; The travel path planning module is used by each AGV to plan a travel path from the starting node to the target node based on the environmental map using an improved path planning algorithm. The AGV displacement module is used for each AGV to move based on the travel path. During the displacement process, the displacement position is corrected by the odometer and gyroscope. The motion information of obstacles is collected by the sensor array, and the travel path is dynamically updated based on the motion information. The displacement information interaction module is used for the intelligent agents of each AGV to interact with each other's displacement information and dynamically update the travel path based on the displacement information until each AGV moves to the target node. The path planning log recording module is used for each AGV to record path planning logs in real time and to store the path planning logs.

7. The intelligent warehousing AGV path planning system based on an improved algorithm as described in claim 6, characterized in that: The environmental data acquisition module is specifically used for: Each AGV obtains the input movement command carrying the target node through the 5G communication module, and controls the sensor array including lidar, camera and infrared sensor based on the movement command to collect environmental data including environmental point cloud data, environmental image data and obstacle distance data.

8. The intelligent warehousing AGV path planning system based on an improved algorithm as described in claim 6, characterized in that: The environment map construction module is specifically used for: Each AGV initializes an environment map based on a two-dimensional grid through a pre-trained and deployed agent. It then performs data fusion analysis on the environmental data to obtain the obstacle positions, sizes, and shapes. Based on the obstacle positions, sizes, and shapes, it marks the two-dimensional grid of the initialized environment map to indicate whether the obstacle is passable.

9. The intelligent warehousing AGV path planning system based on an improved algorithm as described in claim 6, characterized in that: In the path planning module, the path planning algorithm is obtained by improving the heuristic function of the A* algorithm. The formula of the improved heuristic function is: h'(n) = h(n) + k(n); k(n) = α / d(n); Where h'(n) represents the improved heuristic function; h(n) represents the original heuristic function; k(n) represents the obstacle correction factor of the nth node; α represents the adjustment coefficient; and d(n) represents the distance from the nth node to the nearest obstacle. The specific steps of planning the movement path from the starting node to the target node based on the environmental map are as follows: Based on the annotations on the environmental map, candidate nodes for the next displacement of the starting node are selected. An open list and a closed list are created. The starting node and each candidate node are added to the open list, and the estimated total cost of each candidate node is initialized. f(n) = g(n) + h'(n); Where f(n) represents the estimated total cost, which is the estimated total distance or estimated total displacement time from the nth candidate node to the target node; g(n) represents the actual cost from the starting node to the nth candidate node, which is the actual distance or actual displacement time. The candidate node with the lowest estimated total cost is selected from the open list as the current node. It is then determined whether the current node is the target node. If it is, the travel path of each node is recorded to complete path planning. If not, the current node is moved from the open list to the closed list, and the adjacent candidate nodes are expanded for search until the current node is the target node or the open list is empty. The travel path is then recorded.

10. The intelligent warehousing AGV path planning method based on an improved algorithm as described in claim 6, characterized in that: The AGV displacement module is specifically used for: Each AGV moves based on the travel path. During the movement, it acquires the actual position information through the odometer and gyroscope. It compares the actual position information with the travel path to correct the position. It collects motion information of obstacles, including at least the direction and speed of movement, through the sensor array. It sets the obstacle avoidance area based on the motion information and the preset obstacle avoidance distance. It dynamically updates the travel path based on the obstacle avoidance area and the latest environmental map. The displacement information interaction module is specifically used for: Each AGV's intelligent agent interacts with each other through a 5G communication module, including at least the AGV's position, AGV's displacement speed, AGV's displacement direction, AGV's target node, and AGV's task status displacement information. Based on the displacement information, the travel path is dynamically updated until each AGV moves to the target node.

Images