AGV path planning method in industrial scene

Abstract

The application discloses an AGV path planning method in an industrial scene. First, multi-source data is collected in real time, including AGV attribute data, AGV task and cargo attribute data, AGV real-time state data and environment and constraint data; path planning conflict scene identification is performed according to the state of the current AGV, if only a running conflict scene is met, a conventional time window algorithm based on space-time joint planning is used to plan a path for the current AGV; if only a scene constraint conflict scene is met, an enhanced time window algorithm integrating geometric passability constraints is used to plan a path for the current AGV; if neither a running conflict scene nor a scene constraint conflict scene is met, a conventional path planning algorithm is used to plan a path for the current AGV. A hierarchical design strategy of the conventional path planning algorithm, the conventional time window algorithm and the enhanced time window algorithm is adopted to adapt to different scenes, and calculation resource optimization and response speed improvement are realized.

Description

Technical Field

[0001] This invention belongs to the field of AGV scheduling and path planning technology, specifically an AGV path planning method in an industrial scenario. Background Technology

[0002] Automated Guided Vehicles (AGVs) are core equipment for flexible manufacturing and intelligent logistics, and the path planning capability of their scheduling system directly determines the efficiency and stability of the entire operation system. Traditional AGV path planning often uses a single algorithm (such as the A* algorithm or Dijkstra's algorithm), which can only meet the path generation needs in simple scenarios. In complex environments where multiple AGVs work together, it faces a variety of intractable problems.

[0003] At the path planning level, traditional methods are passive in dealing with operational conflicts and lack prevention of scenario-constrained conflicts. Operational conflicts refer to conflicts arising from multiple AGVs competing for path resources during operation (such as intersections, moving in opposite directions, merging, etc.). Existing single planning algorithms mostly focus on spatial path generation, lacking a comprehensive consideration of the temporal dimension. When AGVs encounter movement conflicts such as crossing or moving in opposite directions at intersections or narrow passages, they can often only be resolved passively through emergency stops or reversing, resulting in a significant reduction in operational efficiency. Scenario-constrained conflicts refer to conflicts arising from changes in the effective size of AGVs due to carrying goods, which may lead to scraping, jamming, or collisions in narrow or height-restricted areas. Traditional path planning algorithms do not consider the effective size of AGVs under load and environmental physical constraints (such as path width and clearance height), thus easily causing scraping, jamming, and other interference conflicts with the environment in scenarios such as narrow passages and height-restricted areas.

[0004] Existing methods are lagging in handling congestion-related conflicts. Congestion-related conflicts refer to systemic blockages caused by AGVs becoming abnormally stationary at non-working locations due to malfunctions, power depletion, or other abnormal situations (e.g., AGVs stopping on main roads due to malfunctions, or blocking intersections due to power depletion). Existing methods lack real-time monitoring and early warning mechanisms, typically requiring manual intervention after discovery, which can easily lead to regional or even systemic blockages, causing production or logistics disruptions. Congestion-related conflicts differ from operational conflicts; blockages are physical (AGVs stopping on their paths due to malfunctions), cannot be bypassed by path planning algorithms, and can only be handled manually through alarm mechanisms. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the technical problem this invention aims to solve is to provide an AGV path planning method for industrial scenarios. This method resolves key technical issues in existing AGV path planning, such as passive conflict response, delayed anomaly handling, poor scenario adaptability, and difficulties in expansion and compatibility. It achieves significant improvements in multiple dimensions, including conflict prevention accuracy, system operating efficiency, scenario adaptability, and architecture expansion capabilities, enabling stable and efficient operation of multi-AGV systems in industrial scenarios.

[0006] The present invention solves the aforementioned technical problem by adopting the following technical solution: An AGV path planning method for industrial scenarios, characterized by the following steps: Step 1: Collect multi-source data in real time, including AGV attribute data, AGV task and cargo attribute data, AGV real-time status data, and environmental and constraint data; Step 2: Identify path planning conflict scenarios based on the current AGV status. If only the operational conflict scenario is met, proceed to Step 3; if only the scenario constraint conflict scenario is met, proceed to Step 4; if neither the operational conflict scenario nor the scenario constraint conflict scenario is met, then use a conventional path planning algorithm to plan the path for the current AGV. An operational conflict scenario is considered to be met when the current AGV simultaneously meets the following two conditions; 1) The minimum distance between the estimated path of the current AGV and the planned paths of other AGVs is less than the warning distance threshold; 2) The estimated path of the current AGV overlaps more than the planned path of other AGVs. The scenario constraint conflict scenario is satisfied when the current AGV meets any of the following conditions; 1) The effective size of the current AGV does not meet the physical constraints of the estimated path; 2) The current minimum turning radius of the AGV is greater than the allowable turning radius of any path segment in the estimated path; Step 3: Use a conventional time window algorithm based on spatiotemporal joint planning to plan the path for the current AGV; First, an initial geometric path is generated using a conventional path planning algorithm. Then, based on the AGV kinematic model, the time required for the current AGV to traverse each segment of the initial geometric path is calculated sequentially, thus determining the current AGV occupancy time for each segment. The current AGV occupancy time for each segment is compared one by one with the reserved occupancy time for the corresponding segment in the global time window. If the current AGV occupancy time overlaps with the reserved occupancy time, the segment is considered a conflict segment, and conflict resolution is performed to obtain the planned path. Finally, the geometric passability of the planned path is verified. If the effective size or minimum turning radius of the current AGV does not meet the physical constraints of the planned path, step four is executed, and path planning is re-performed; otherwise, the planned path is the final planned path for the current AGV. Step 4: Use an enhanced time window algorithm with integrated geometric passability constraints to plan the path for the current AGV; First, considering a safety margin based on the effective size of the AGV, a geometric passability check is performed on all path segments in the environmental map according to the effective size of the AGV with the safety margin. Path segments that pass the geometric passability check are taken as feasible path segments, thus obtaining a feasible path subnet. The geometric passability check includes: 1) Width passability check: If the width of the path segment is greater than or equal to the effective width of the current AGV considering the safety margin, the width passability check is successful; otherwise, the width passability check is unsuccessful. 2) Height passability check: If the net height of the path segment is greater than or equal to the effective height of the current AGV considering the safety margin, the height passability check is successful; otherwise, the height passability check is unsuccessful. 3) Turning radius passability check: If the allowable turning radius of the path segment is greater than or equal to the minimum turning radius of the current AGV, the turning radius passability check is successful; otherwise, the turning radius passability check is unsuccessful. Then, based on the feasible path subnet, the conventional time window algorithm based on spatiotemporal joint planning in the third step is used for path planning.

[0007] Furthermore, in the third step, the following three strategies are used sequentially to resolve the conflict: Strategy 1: Insert a waiting time to postpone the entry time; Insert a waiting time before the current AGV enters the conflict path segment. The waiting time is equal to the difference between the reservation end time of the conflict path segment and the original planned time for the current AGV to arrive at the conflict path segment. The insertion position is the end position of the previous path segment of the conflict path segment or the safe waiting area before the conflict path segment, and ensures that the actual time of the current AGV to arrive at the task termination point is earlier than or equal to the task deadline time. Strategy 2: Fine-tune driving speed; If the current AGV's start time for occupying is earlier than the scheduled end time of the conflict path segment, and the time difference is less than or equal to the time difference threshold, then reduce the current AGV's speed before the conflict path segment, so that the start time for occupying is postponed until after the scheduled end time of the conflict path segment; if the time difference is greater than the time difference threshold or if there is still a conflict after speed fine-tuning, then try strategy three. Strategy 3: Locally adjust the initial geometric path; A conventional path planning algorithm is used to generate a new local path in the local area before the conflicting path segment. The time period occupied by the new local path is calculated and time-dimensional conflict detection and conflict resolution are performed until a conflict-free planned path is generated.

[0008] Furthermore, in the second step, the shortest distance between any path segment in the current AGV's estimated path and all path segments in other AGVs' planned paths is calculated. If two path segments intersect or are adjacent in space, the shortest distance between the path pairs is zero; if two path segments do not intersect, the shortest distance between any two points on the two path segments is calculated, and the minimum of the shortest distances between all point pairs is taken as the shortest distance between the path segment pairs. All path segments in the current AGV's estimated path are traversed, and the minimum of the shortest distances between all path segment pairs is taken as the minimum distance between the current AGV's estimated path and other AGVs' planned paths.

[0009] Furthermore, in the second step, it is determined whether any path segment in the current AGV's estimated path overlaps with path segments in other AGV's planned paths. If the identifiers of two path segments are the same, or if two path segments completely overlap in space, they are determined to be overlapping path segments. All path segments in the current AGV's estimated path are traversed to obtain all overlapping path segments. The sum of the lengths of all overlapping path segments is divided by the total length of the current AGV's estimated path to obtain the path overlap degree.

[0010] Furthermore, based on the task start and end points and the environment map, a conventional path planning algorithm is used to generate the estimated path for the current AGV; the conventional path planning algorithm is either the A* algorithm or the Dijkstra algorithm.

[0011] Furthermore, the environmental map needs to consider path segments containing congestion sources; for a stationary AGV, if the current position is a non-preset working point of the stationary AGV, and the cumulative stagnation time at the non-preset working point is greater than the maximum allowable stagnation time at the non-preset working point, then the stationary AGV is a congestion source.

[0012] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention adopts a hierarchical design strategy of "conventional path planning algorithm → conventional time window algorithm → enhanced time window algorithm". Conventional path planning algorithms (such as A* algorithm and Dijkstra's algorithm) are suitable for conflict-free scenarios, with low computational complexity and fast response speed. Conventional time window algorithms prevent runtime conflicts at the start of the task through spatiotemporal joint planning, and are suitable for scenarios with only runtime conflicts, with low computational resource consumption and high execution efficiency. Enhanced time window algorithms avoid scenario constraint conflicts at the start of the task through geometric passability preprocessing and spatiotemporal joint planning, and are suitable for scenarios with scenario constraint conflicts or simultaneous runtime and scenario constraint conflicts. The hierarchical design strategy mainly achieves the following: First, computational resource optimization; conventional path planning algorithms only require simple path search and do not require time window reservation and conflict resolution; conventional time window algorithms require time window reservation and conflict resolution, while enhanced time window algorithms add geometric passability verification on the basis of conventional time window algorithms. Therefore, using conventional path planning algorithms in conflict-free scenarios can save about 80% of computational resources. Second, improved response speed: Conventional path planning algorithms typically have response times in the millisecond range, conventional time window algorithms in the hundreds of milliseconds range, and enhanced time window algorithms in the seconds range. The most suitable algorithm can be selected based on the complexity of the scenario to plan the path, avoiding overcomputation and improving system response speed. Third, system load balancing: In multi-AGV systems, many simple tasks use conventional path planning algorithms, medium-complexity tasks use conventional time window algorithms, and complex tasks use enhanced time window algorithms. Intelligent strategy matching achieves reasonable distribution of system load, avoiding system overload caused by using the most complex algorithm for all tasks. Fourth, scalability: The layered design facilitates the subsequent expansion of new strategy types (such as dynamic obstacle avoidance strategies, multi-task priority scheduling strategies, etc.) without modifying the core architecture; only new judgment conditions need to be added to the strategy matching rules. Therefore, this invention, through an intelligent strategy matching mechanism, automatically selects a matching algorithm for path planning based on the scene recognition results, ensuring conflict resolution while achieving optimal allocation of computing resources and maximizing system performance. In real-world testing of multi-AGV scheduling scenarios in warehousing and logistics, this invention reduces the overall conflict rate from 18% in existing technologies to below 2.5%. Compared to a single planning algorithm, the system's operating efficiency is improved by more than 25%, while intelligent selection through the algorithm reduces the system's computational resource consumption by approximately 30%.

[0013] 2. A conventional time window algorithm based on spatiotemporal joint planning achieves dual spatiotemporal reservation of path resources (planning paths in the spatial dimension and reserving path occupancy time periods in the temporal dimension, realizing joint spatial and temporal planning). Combined with the AGV kinematic model, the occupancy time periods of path segments are accurately calculated, enhancing the ability to prevent operational conflicts and avoiding the problem of "feasible paths but time conflicts" in traditional single spatial planning. Compared with planning methods that rely solely on geometric paths, this invention can resolve more than 95% of operational conflicts such as intersections and oncoming traffic in advance, reducing inefficient operations such as AGV emergency stops and reversing, and shortening the average completion time of a single AGV task by about 12%.

[0014] 3. The enhanced time window algorithm integrating geometric passability constraints dynamically selects feasible paths based on the effective size of the AGV through geometric passability verification, completely solving the pain point of "AGV scraping and getting stuck after being loaded" in traditional planning. In tests on scenarios with variable cargo sizes, such as electronic component warehousing, this strategy reduced the incidence of environmental interference accidents from 9% in existing technologies to nearly 0, and reduced equipment maintenance costs by approximately 45% compared to planning schemes that do not consider cargo size.

[0015] 4. Before path planning, a dual-judgment logic of "position verification + timeout timing" continuously monitors and identifies blockage sources, feeding back blockage status information to the environmental map in real time, thus integrating it into path planning and preventing the planned path from passing through blocked areas. Compared to the traditional method of relying on manual inspection, alarm response time is reduced from minutes to seconds, and the overall downtime caused by blockages is reduced by approximately 90%. In AGV intensive operation scenarios, equipment utilization is increased by more than 15%. Attached Figure Description

[0016] Figure 1 This is an overall flowchart of the present invention; Figure 2 It is a path planning graph obtained using a conventional path planning algorithm in a scenario of operational conflict; Figure 3 It is a path planning graph obtained by using a conventional time window algorithm based on spatiotemporal joint planning in the context of operational conflict scenarios; Figure 4 It is a path planning graph obtained using a conventional path planning algorithm under scenario-constrained conflict. Figure 5 It is a path planning graph obtained by using an enhanced time window algorithm with integrated geometric passability constraints in scenario-constrained conflict scenarios. Detailed Implementation

[0017] Specific embodiments are given below with reference to the accompanying drawings. These specific embodiments are only used to describe the technical solution of the present invention in detail and are not intended to limit the scope of protection of this application.

[0018] like Figure 1 As shown, this invention provides an AGV path planning method in an industrial setting, comprising the following steps: Step 1: Collect multi-source data in real time through a multi-interface collaborative acquisition mechanism, including AGV attribute data, AGV task and cargo attribute data, AGV real-time status data, and environmental and constraint data; The AGV attribute data is obtained through the direct connection interface of the AGV vehicle controller, including vehicle identifier (used for AGV identification, each AGV has a unique vehicle identifier), AGV dimensions (including length, width and height), motion characteristic parameters (including drive type (differential drive and omnidirectional movement), minimum turning radius, maximum speed, maximum acceleration and maximum deceleration) and load parameters (including maximum load). The AGV task and cargo attribute data are obtained through the task distribution interface of the central scheduling system, including task identifier (for task identification), task priority, task deadline, task start and end point pose (including coordinate position and AGV orientation angle), cargo status (including loaded and empty status) and cargo dimensions (including length, width and height). The AGV's real-time status data is received in real time through the IoT gateway, including the current pose (containing the AGV's coordinate position and orientation angle, used to determine the starting point of path planning at the start of the task, usually consistent with the pose of the task's starting point), real-time speed, downtime, and the identifier of the current path segment. The environmental and constraint data is read from the scheduling system database, including an environmental map (containing path segment identifiers and coordinate mappings), path segment width and clearance height (i.e., the maximum height allowed to pass above the path segment), speed limits, allowed turning radius of the path segment, docking station coordinates, and work point coordinates. The environmental map needs to consider congested path segments, i.e., path segments containing congestion sources. For a stationary AGV, if the current position is the preset work point of the stationary AGV, then the stationary AGV is not a congestion source. If the current position is a non-preset work point of the stationary AGV, and the cumulative dwell time of the stationary AGV at the non-preset work point is greater than the maximum allowable dwell time at the non-preset work point, then the stationary AGV is a congestion source. The above data covers the full range of AGV operation and conflict prevention needs, providing complete and accurate data support for subsequent scenario identification and strategy matching.

[0019] Step 2: Identify path planning conflict scenarios based on the current AGV status. If the current AGV only meets the operational conflict scenario, proceed to Step 3, which uses a conventional time window algorithm based on spatiotemporal joint planning for path planning. If the current AGV only meets the scenario constraint conflict scenario (including simultaneously meeting both scenarios, or only meeting the scenario constraint conflict scenario), proceed to Step 4, which uses an enhanced time window algorithm integrating geometric throughput constraints for path planning. If the current AGV does not meet either the operational conflict scenario or the scenario constraint conflict scenario, then a conventional path planning algorithm (such as the A* algorithm or Dijkstra's algorithm) is used for path planning. If multiple AGVs need to plan paths at the same time, they are queued according to task priority, with AGVs with higher task priority being planned first. Path planning conflict scenarios are divided into two categories: operational conflict scenarios and scenario constraint conflict scenarios. An AGV is considered to meet an operational conflict scenario when it meets both of the following conditions. 1) The minimum distance between the estimated path of the current AGV and the planned paths of other AGVs is less than the warning distance threshold; 2) The current AGV's estimated path overlaps more than the path overlap threshold with the other AGV's planned path.

[0020] The current AGV's predicted path uses a conventional path planning algorithm, which performs a simple path search based on the task's start and end points and the environment map, generating a collision-free path that only considers geometric feasibility. The minimum distance is used to determine the spatial proximity between paths. The specific calculation process is as follows: For any path segment in the current AGV's estimated path, calculate the shortest distance between it and all path segments in other AGV's planned paths. If two path segments intersect or are adjacent in space, the shortest distance between the path pairs is zero; if two path segments do not intersect, calculate the shortest distance between any two points on the two path segments, and take the minimum of the shortest distances between all point pairs as the shortest distance between the path segments; traverse all path segments in the current AGV's estimated path, and take the minimum of the shortest distances between all path segment pairs as the minimum distance between the current AGV's estimated path and other AGV's planned paths.

[0021] Path overlap is used to determine the degree of overlap between paths. The specific calculation process is as follows: For any path segment in the current AGV's estimated path, determine whether it overlaps with path segments in other AGV's planned paths. If the identifiers of the two path segments are the same, or the two path segments completely overlap in space (i.e., the start and end coordinates of the two path segments are the same), then they are determined to be overlapping path segments. Traverse all path segments in the current AGV's estimated path to obtain all overlapping path segments. Divide the sum of the lengths of all overlapping path segments by the total length of the current AGV's estimated path to obtain the path overlap.

[0022] The AGV is deemed to satisfy the scenario constraint conflict scenario if it meets any of the following conditions: 1) The effective dimensions of the current AGV do not meet the physical constraints of the estimated path, that is, the effective width of the current AGV is greater than the width of any path segment in the estimated path, or / and the effective height is greater than the net height of any path segment in the estimated path. 2) The current minimum turning radius of the AGV is greater than the allowable turning radius of any path segment in the estimated path.

[0023] When loaded, the effective size of the AGV is the actual size of the AGV after loading goods; when unloaded, the effective size of the AGV is the size of the AGV itself.

[0024] Step 3: Use a conventional time window algorithm based on spatiotemporal joint planning for path planning; 3.1 Generate the initial geometric path Based on the start and end points of the task and the environment map, a conventional path planning algorithm is used to generate an initial geometric path that only considers geometric feasibility and is collision-free. The initial geometric path only considers the spatial dimension to ensure that the AGV can connect from the start point to the end point, but does not consider the conflict problem in the time dimension. 3.2 Time-dimension conflict detection The path planner simulates the current AGV's journey along the initial geometric path. Based on the AGV's kinematic model, it calculates the time required for the current AGV to pass through each path segment in sequence, thereby determining the current AGV's occupation time for each path segment in the initial geometric path. The start time of occupation for the first path segment is equal to the task start time, and the start time of occupation for the next path segment is equal to the end time of occupation for the previous path segment. The current AGV occupancy period of each path segment in the initial geometric path is compared with the reserved occupancy period of the corresponding path segment in the global time window (i.e., the reservation record table of occupancy periods of all path segments in the environment map). If the current AGV occupancy period of the path segment overlaps with the reserved occupancy period of the same path segment, then there is a time dimension conflict between the current AGV occupancy period and the reserved occupancy period of the path segment, which is a conflicting path segment. 3.3 Spatiotemporal Conflict Resolution When there is a time conflict between the current AGV's occupied time period and the reserved time period on a path segment, the path planner does not directly discard the initial geometric path, but uses a conflict resolution strategy to resolve the spatiotemporal conflict and obtain the planned path for the current AGV. There are three conflict resolution strategies, which are tried in sequence. First, strategy one is tried. If it cannot be resolved, strategy two is tried. If it still cannot be resolved, strategy three is tried. Strategy 1: Insert a waiting time to postpone the entry time; For conflicting path segments, a waiting time is inserted before the current AGV enters the conflicting path segment. The waiting time is equal to the difference between the reservation end time of the conflicting path segment and the original planned time for the current AGV to arrive at the conflicting path segment. The waiting time is inserted at the end of the previous path segment or the safe waiting area before the conflicting path segment, so that the current AGV waits before entering the conflicting path segment and enters after the conflicting path segment is released. Since increasing the waiting time will affect the task completion time, it is necessary to ensure that the actual time of the current AGV arriving at the task termination point is earlier than or equal to the task deadline.

[0025] Strategy 2: Fine-tune driving speed to stagger time windows; For conflicting path segments, if the current AGV's start time for occupying the conflicting path segment is earlier than the scheduled end time for occupying the conflicting path segment, and the time difference is less than or equal to the time difference threshold, then the current AGV's speed before the conflicting path segment is reduced to extend the time to reach the conflicting path segment, thus delaying the start time for occupying the conflicting path segment until after the scheduled end time for occupying the conflicting path segment; if the time difference is greater than the time difference threshold or if the conflict still exists after speed fine-tuning, then strategy three is attempted.

[0026] Strategy 3: Locally adjust the initial geometric path to bypass conflicting path segments; Find alternative path segments before the conflicting path segments. That is, use conventional path planning algorithms to generate new local paths in the local area before the conflicting path segments to bypass the conflicting path segments. Calculate the time period occupied by the new local paths and perform time-dimensional conflict detection and spatiotemporal conflict resolution until a conflict-free planned path is generated. 3.4 Perform geometric passability verification on the planned path; If the effective size or minimum turning radius of the current AGV does not meet the physical constraints of the planned path, that is, the effective width of the current AGV is greater than the width of the path segment, or the effective height is greater than the net height of the path segment, or the minimum turning radius of the current AGV is greater than the allowable turning radius of the path segment, it indicates that the geometric passability verification of the planned path is unsuccessful. Then, the fourth step is executed, and the enhanced time window algorithm with integrated geometric passability constraints is used to re-plan the path. If the effective size or minimum turning radius of the current AGV meets the physical constraints of the planned path, it indicates that the geometric passability verification of the planned path is successful, and the planned path is the final planned path of the current AGV.

[0027] Figure 2 , 3The images show the path planning diagrams obtained by the conventional path planning algorithm and the conventional time window algorithm based on spatiotemporal joint planning, respectively. AGV1 starts from the starting point S1 and travels to the destination E1 via nodes P1-P7. AGV2 starts from the starting point S2 and travels to the destination E2 via nodes P8, P9, P3, P4, P10, and P11. AGV1 is planned first, followed by AGV2. When using the conventional path planning algorithm, the planned paths of the two vehicles overlap in the path segment P3-P4. Due to the lack of time dimension reservation, there is a time occupation conflict between the two vehicles in this path segment. Operational conflicts such as intersections or head-on conflicts are likely to occur in the path segment P3-P4, which need to be resolved by emergency stopping or reversing. By adopting the conventional time window algorithm based on spatiotemporal joint planning, the path planner reserves the time slots for each path segment in the time dimension. When AGV2 detects a time conflict with AGV1 in the P3-P4 path segment during planning, it enters the path segment only after AGV1 has finished occupying the time slot, thus achieving dual time and space reservation. The two vehicles pass through the P3-P4 path segment in the order of the time window, avoiding operational conflicts and resolving conflicts without the need for emergency stopping or reversing.

[0028] Step 4: Path planning is performed using an enhanced time window algorithm that integrates geometric passability constraints; 4.1 Screen the path segments of the environmental map to obtain feasible path subnets; Based on the effective size of the AGV, a safety margin is considered, and a geometric passability check is performed on all path segments in the environmental map according to the current effective size of the AGV with the safety margin, including: 1) Width passability check: If the width of the path segment is greater than or equal to the effective width of the current AGV considering the safety margin, it indicates that the path segment is passable in the width direction, and the width passability check is successful; otherwise, the width passability check is unsuccessful. 2) Height passability check: If the clearance height of the path segment is greater than or equal to the effective height of the current AGV considering the safety margin, it indicates that the path segment is passable in the height direction, and the height passability check is successful; otherwise, the height passability check is unsuccessful. 3) Turning radius passability check: If the allowable turning radius of the path segment is greater than or equal to the minimum turning radius of the current AGV, it means that the path segment is passable within the turning radius, and the turning radius passability check is successful; otherwise, the turning radius passability check is unsuccessful; since omnidirectional AGVs turn in place, they usually do not perform turning radius passability checks. The path segments that pass the geometric passability check are considered as feasible path segments, and all feasible path segments form a feasible path subnet. 4.2 Based on the feasible path subnet, the conventional time window algorithm based on spatiotemporal joint planning is used to plan the path of the current AGV. For details, please refer to step 3.

[0029] Figure 4 , 5 The images show path planning diagrams obtained by a conventional path planning algorithm and an enhanced time window algorithm integrating geometric passability constraints, respectively. An AGV carrying goods travels from the starting point S (node ​​P1) to the ending point E (node ​​P5). The path planning only considers the geometric shortest path, selecting a straight path P1→P2→P3→P4→P5. However, the width or clearance height of the P2-P4 path segment is less than the effective width and effective height of the AGV under load, failing to meet the geometric passability physical constraints. When the AGV travels along this path, it is prone to scraping, jamming, or being unable to pass through narrow path segments, thus violating scenario constraints. When using the enhanced time window algorithm with integrated geometric passability constraints, the path planner performs geometric passability verification on the environmental map based on the effective dimensions of the AGV under load. It determines that the path segment P2-P4 in the straight path P1→P2→P3→P4→P5 does not meet the width or clearance height constraints, and removes it. The path is then replanned in the feasible path subnet, resulting in the path P1→P6→P7→P8→P9→P10→P5. This path meets the geometric passability verification, and the AGV can pass safely, avoiding environmental interference conflicts such as scratches and jamming.

[0030] Any aspects not covered in this invention are applicable to existing technologies.

Claims

1. An AGV path planning method in an industrial setting, characterized in that, Includes the following steps: Step 1: Collect multi-source data in real time, including AGV attribute data, AGV task and cargo attribute data, AGV real-time status data, and environmental and constraint data; Step 2: Identify path planning conflict scenarios based on the current AGV status. If only the operational conflict scenario is met, proceed to Step 3; if only the scenario constraint conflict scenario is met, proceed to Step 4; if neither the operational conflict scenario nor the scenario constraint conflict scenario is met, then use a conventional path planning algorithm to plan the path for the current AGV. An operational conflict scenario is considered to be met when the current AGV simultaneously meets the following two conditions; 1) The minimum distance between the estimated path of the current AGV and the planned paths of other AGVs is less than the warning distance threshold; 2) The estimated path of the current AGV overlaps more than the planned path of other AGVs. The scenario constraint conflict scenario is satisfied when the current AGV meets any of the following conditions; 1) The effective size of the current AGV does not meet the physical constraints of the estimated path; 2) The current minimum turning radius of the AGV is greater than the allowable turning radius of any path segment in the estimated path; Step 3: Use a conventional time window algorithm based on spatiotemporal joint planning to plan the path for the current AGV; First, an initial geometric path is generated using a conventional path planning algorithm. Then, based on the AGV kinematic model, the time required for the current AGV to traverse each segment of the initial geometric path is calculated sequentially, thus determining the current AGV occupancy time for each segment. The current AGV occupancy time for each segment is compared one by one with the reserved occupancy time for the corresponding segment in the global time window. If the current AGV occupancy time overlaps with the reserved occupancy time, the segment is considered a conflict segment, and conflict resolution is performed to obtain the planned path. Finally, the geometric passability of the planned path is verified. If the effective size or minimum turning radius of the current AGV does not meet the physical constraints of the planned path, step four is executed, and path planning is re-performed; otherwise, the planned path is the final planned path for the current AGV. Step 4: Use an enhanced time window algorithm with integrated geometric passability constraints to plan the path for the current AGV; First, considering a safety margin based on the effective size of the AGV, a geometric passability check is performed on all path segments in the environmental map according to the effective size of the AGV with the safety margin. Path segments that pass the geometric passability check are taken as feasible path segments, thus obtaining a feasible path subnet. The geometric passability check includes: 1) Width passability check: If the width of the path segment is greater than or equal to the effective width of the current AGV considering the safety margin, the width passability check is successful; otherwise, the width passability check is unsuccessful. 2) Height passability check: If the net height of the path segment is greater than or equal to the effective height of the current AGV considering the safety margin, the height passability check is successful; otherwise, the height passability check is unsuccessful. 3) Turning radius passability check: If the allowable turning radius of the path segment is greater than or equal to the minimum turning radius of the current AGV, the turning radius passability check is successful; otherwise, the turning radius passability check is unsuccessful. Then, based on the feasible path subnet, the conventional time window algorithm based on spatiotemporal joint planning in the third step is used for path planning.

2. The AGV path planning method in industrial scenarios according to claim 1, characterized in that, In the third step, the following three strategies are used in sequence to resolve the conflict: Strategy 1: Insert a waiting time to postpone the entry time; Insert a waiting time before the current AGV enters the conflict path segment. The waiting time is equal to the difference between the reservation end time of the conflict path segment and the original planned time for the current AGV to arrive at the conflict path segment. The insertion position is the end position of the previous path segment of the conflict path segment or the safe waiting area before the conflict path segment, and ensures that the actual time of the current AGV to arrive at the task termination point is earlier than or equal to the task deadline time. Strategy 2: Fine-tune driving speed; If the current AGV's start time for occupying is earlier than the scheduled end time of the conflict path segment, and the time difference is less than or equal to the time difference threshold, then reduce the current AGV's speed before the conflict path segment, so that the start time for occupying is postponed until after the scheduled end time of the conflict path segment; if the time difference is greater than the time difference threshold or if there is still a conflict after speed fine-tuning, then try strategy three. Strategy 3: Locally adjust the initial geometric path; A conventional path planning algorithm is used to generate a new local path in the local area before the conflicting path segment. The time period occupied by the new local path is calculated and time-dimensional conflict detection and conflict resolution are performed until a conflict-free planned path is generated.

3. The AGV path planning method in industrial scenarios according to claim 1, characterized in that, In the second step, the shortest distance between any path segment in the current AGV's estimated path and all path segments in other AGVs' planned paths is calculated. If two path segments intersect or are adjacent in space, the shortest distance between the path pairs is zero. If two path segments do not intersect, the shortest distance between any two points on the two path segments is calculated, and the minimum of the shortest distances between all point pairs is taken as the shortest distance between the path segment pairs. All path segments in the current AGV's estimated path are traversed, and the minimum of the shortest distances between all path segment pairs is taken as the minimum distance between the current AGV's estimated path and other AGVs' planned paths.

4. The AGV path planning method in industrial scenarios according to claim 1, characterized in that, In the second step, it is determined whether any path segment in the current AGV's estimated path overlaps with path segments in other AGV's planned paths. If two path segments have the same identifier or completely overlap in space, they are determined to be overlapping path segments. All path segments in the current AGV's estimated path are traversed to obtain all overlapping path segments. The sum of the lengths of all overlapping path segments is divided by the total length of the current AGV's estimated path to obtain the path overlap degree.

5. The AGV path planning method in industrial scenarios according to any one of claims 1 to 4, characterized in that, Based on the task start and end points and the environment map, the estimated path of the current AGV is generated using a conventional path planning algorithm; the conventional path planning algorithm is either the A* algorithm or the Dijkstra algorithm.

6. The AGV path planning method in industrial scenarios according to claim 5, characterized in that, The environment map needs to consider path segments containing congestion sources; for a stationary AGV, if the current position is a non-preset working point of the stationary AGV, and the cumulative stagnation time at the non-preset working point is greater than the maximum allowable stagnation time at the non-preset working point, then the stationary AGV is a congestion source.

Images