Multi-AGV cooperative path planning method and system for rim production workshop

Abstract

The invention provides a multi-AGV cooperative path planning method and system for a rim production workshop, and belongs to the technical field of path planning of robots. The method comprises the following steps: firstly, modeling a rim production workshop environment; mapping the planned path of the AGV to a corridor topology set, calculating and managing the occupancy state of each corridor section in the corridor topology set on a time axis based on the driving state of the AGV, and generating an occupancy state table; based on the occupancy state table, path planning is carried out by adopting an improved TRTD-RRT algorithm; carrying out space and time dimension conflict detection on each generated AGV path; if the strong conflict is detected, starting a cooperative scheduling strategy, and adjusting the passing priority of the AGV or performing local path re-planning according to a preset rule; and converting the finally determined path into a motion control instruction of the AGV, and performing closed-loop control based on feedback of the sensor. According to the invention, pre-avoidance and active cooperation of path conflicts are realized, so that the stability and operation efficiency of path planning and the safety of the whole system are improved.

Description

Technical Field

[0001] This invention relates to a multi-AGV collaborative path planning method and system for wheel rim production workshops, belonging to the field of robot path planning technology. Background Technology

[0002] In industrial manufacturing, AGVs are used for transporting semi-finished and finished products. With increasing automation in workshops, the collaborative operation of multiple AGVs has become commonplace. Compared to typical industrial scenarios, wheel rim workshops are characterized by dense equipment, narrow passageways, and limited turning radii. Furthermore, they have numerous material nodes and fixed task cycles, placing higher demands on the real-time performance and stability of AGV path planning. While some existing path planning methods can generate feasible paths in static environments, they primarily cater to single-AGV operation scenarios. For scenarios with multiple AGVs operating simultaneously, traditional technologies typically employ local obstacle avoidance, priority allocation, or centralized scheduling to resolve path conflicts. This often necessitates path correction after a conflict occurs, leading to repeated replanning, localized congestion, and decreased operational efficiency. Moreover, existing solutions usually only consider the real-time positions of other AGVs without incorporating their planned paths as constraints in the path generation phase. This results in a lack of foresight in path decision-making, hindering the formation of system-level collaborative effects. Therefore, in the unique production environment of wheel rim workshops, a method is needed that can proactively utilize multi-AGV path information and achieve collaborative planning to improve path continuity and overall operational efficiency.

[0003] To enhance the intelligence level of material transportation in workshops, some enterprises have begun to introduce AGV global path planning and obstacle avoidance control technology to achieve automatic generation of AGV travel routes and collision avoidance. Existing solutions mostly employ planning methods based on global maps, using preset workshop topology and operating trajectories to optimize AGV paths between equipment areas, channel nodes, and workstations. Some technologies utilize improved AGV systems... The algorithm or RRT algorithm calculates the shortest path, enabling the AGV to autonomously reach the designated location in the workshop environment. It also uses lidar or ultrasonic sensors to detect obstacles during operation, thereby achieving path correction and driving safety control.

[0004] In the process of implementing the embodiments of this disclosure, at least the following problems were found in the related art: Existing path planning algorithms are mostly designed for standardized spatial designs, making them ill-suited to the real-world environment of wheel rim workshops, characterized by dense equipment, narrow passageways, and dynamically changing layouts. Furthermore, they are sensitive to changes in workshop layout and production status, struggling to cope with dynamic adjustments to the wheel rim workshop environment and real-time logistical disruptions. Traditional path planning algorithms, such as A... Algorithms often require high computational overhead in complex two-dimensional environments; although the RRT algorithm can quickly generate feasible paths, the results are often not optimal and converge slowly in areas with dense obstacles; it does not consider the information association and conflict handling between multiple AGVs, and cannot achieve global path planning and safe scheduling. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-AGV collaborative path planning method and system for wheel rim production workshops. By introducing spatiotemporal occupancy constraint modeling based on passage sections and integrating occupancy risk perception into the core path planning algorithm, the method achieves proactive avoidance and coordination of path conflicts, thereby improving the stability, operational efficiency, and overall system security of path planning.

[0006] To achieve the above objectives, the present invention employs the following technical solution: A multi-AGV collaborative path planning method for a wheel rim production workshop includes the following steps: Model the environment of the wheel rim production workshop to obtain a grid map, corridor topology set, and narrow resource segment set; The planned path of the AGV is mapped to the corridor topology set, and the occupancy status of each corridor segment in the corridor topology set on the time axis is calculated and managed based on the driving status of the AGV, generating an occupancy status table. Based on the occupancy status table, an improved TRTD-RRT algorithm is used for path planning. When expanding and reconnecting path nodes, the improved TRTD-RRT algorithm takes the occupancy risk factor of the target corridor segment as part of the comprehensive cost function. Perform spatial and temporal conflict detection on each generated AGV path; If a strong conflict is detected, a collaborative scheduling strategy is activated, adjusting the AGV's passage priority or performing local path replanning according to preset rules. The finalized path is converted into motion control commands for the AGV, and closed-loop control is performed based on sensor feedback.

[0007] Preferably, the modeling of the wheel rim production workshop environment specifically includes: The system collects environmental data from the workshop using sensors, generates a grid map, and inflates the obstacle areas. Extract passage segments that meet preset width, length, and connectivity constraints from the processed map to form a corridor topology set; Based on the preset minimum passage width threshold for AGVs, a set of narrow resource segments is identified and marked from the corridor topology set.

[0008] Preferably, the occupancy status table is used to record the occupancy status of each corridor segment by different AGVs in different time intervals; Based on the AGV's travel distance and preset speed within the corridor section, its entry and exit times are calculated to determine the occupied time interval.

[0009] Preferably, the improved TRTD-RRT algorithm employs a bidirectional expanding search tree framework, and its improvements include: When a new path node is added to the search tree, a local rewiring is performed, and the comprehensive cost function is used to compare the path node with its neighborhood radius. The connection relationships of the nodes within the node are adjusted, and nodes with lower overall cost functions are selected for connection; The new path nodes are selected based on a dynamic target bias sampling mechanism that considers the occupancy risk factor of the corridor passage segment in the target direction: Calculate the dynamic target bias sampling probability and generate random numbers. If random number If the probability is less than the dynamic target bias sampling probability, then target bias sampling is performed, and a new path node is determined by extending the path towards the target point by a fixed step size; otherwise, a new path node is randomly selected in free space.

[0010] Preferably, the comprehensive cost function is used to compare the path node with its neighborhood radius. The connection relationships between nodes within the node are adjusted as follows: After generating new path nodes, use this as the center and the neighborhood radius The search term is defined as the range within which all existing nodes within the region are considered as candidate nodes for evaluation. Calculate the combined cost between the new path node and the node to be determined, and select the node with the lower combined cost as the target node for connection.

[0011] Preferably, the formula for calculating the dynamic target bias sampling probability is: , in, For dynamic target bias sampling probability. To minimize the search probability, To maximize the search probability, For the maximum iteration count, For the current iteration count, To control the growth rate, The occupancy risk factor for the corridor segment in the target direction is calculated using the following formula: , in, For a narrow set of resource segments, For width parameter, For width threshold, This is the attenuation sensitivity coefficient.

[0012] Preferably, the comprehensive cost function is as follows: , in, This represents the total cost from the current node to the parent node. As the parent node, For the current node, This represents the cumulative cost from the current node to the parent node. This represents the Euclidean distance from the current node to its parent node. The weighting coefficients are used to balance length and smoothness. As a risk cost weight, The cost of curvature or angle for connecting branches.

[0013] Preferably, the collision detection includes: Spatial conflict detection: Determine whether the projected areas of the planned paths of different AGVs overlap on the corridor section; Time conflict detection: Determine whether there is an overlap in the time intervals during which different AGVs occupy the same corridor segment; If both spatial overlap and temporal intersection exist simultaneously, it is considered a strong conflict.

[0014] Preferably, the collaborative scheduling strategy includes: The passage priority of each conflicting AGV is calculated based on its remaining path length and estimated planning time, using the following formula: , in, For the first The remaining path length of the AGV from the target point. For the first The estimated planning time for each AGV is as follows: As a length-sensitive factor, As a time-sensitive factor, For the first Priority of each AGV.

[0015] If a high-priority AGV has a strong conflict with a low-priority AGV, the high-priority AGV will pass first, while the low-priority AGV will either wait or trigger a local path replanning strategy.

[0016] A multi-AGV collaborative path planning system for a wheel rim production workshop, the system comprising: The environment modeling module is used to build a workshop environment model and output a workshop map, a corridor topology set, and a narrow resource segment set. The corridor occupancy management module is used to map AGV paths to corridor topology and manage the time occupancy status of each corridor segment. The path planning module is used to plan the path for AGVs based on the occupancy status and using the TRTD-RRT algorithm. The conflict detection module is used to detect spatial and temporal conflicts between different AGV paths; The collaborative scheduling module is used to make scheduling decisions for conflicting AGVs when strong conflicts are detected. The execution control module is used to convert the planned path into AGV motion commands and realize closed-loop control.

[0017] The advantages of this invention are as follows: By abstracting the passage area into occupiable path resources and introducing an occupancy constraint mechanism during the path planning stage, this invention achieves early avoidance of potential path conflicts among multiple AGVs, reducing the probability of oncoming intersections and local congestion. The TRTD-RRT algorithm combines topology-based local rewiring based on passage section occupancy risk perception and a dynamic target bias sampling mechanism, effectively improving path generation efficiency and path quality. Through a linkage mechanism of environment modeling, path planning, conflict detection, and collaborative scheduling, the system avoids resource contention among multiple AGVs and traffic deadlock, ensuring operational continuity and stability, and significantly improving operational safety. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0019] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0020] Figure 2 This is a schematic diagram of the system flow of the present invention.

[0021] Figure 3 This is a schematic diagram of the improved TRTD-RRT algorithm of this invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Traditional path planning methods often encounter problems such as path conflicts, misjudgments, traffic congestion, redundant scheduling, and excessively long planning times when operating multiple AGVs, leading to reduced logistics efficiency and even safety risks. Existing multi-AGV path planning technologies lack system coordination capabilities, dynamic prediction capabilities, and global optimization capabilities in the wheel rim workshop environment. This invention addresses the collaborative operation requirements of multiple AGVs (Automated Guided Vehicles) in wheel rim production workshops by introducing a "time and space occupancy constraint modeling based on passage sections" approach to avoid redundant replanning. It constructs an overall architecture for an AGV path planning system based on a "planned path occupancy corridor" mechanism.

[0024] First, the fixed structure of the production workshop is topologically abstracted, forming a narrow area around the current feasible path, and a map model centered on "corridors" is generated. The passage cost of the area is determined based on the corridor width and equipment distribution. Then, based on the planned paths of different AGVs, the occupancy status of the corridor segment is recorded on the time axis, and the occupancy table is continuously updated for occupancy prediction before path planning. Planning is performed based on this occupancy status, prioritizing unoccupied routes or routes with low occupancy conflicts. An improved TRTD-RRT algorithm is proposed by introducing a topology path rerouting mechanism and a dynamic target bias sampling mechanism based on the risk perception of passage segment occupancy. If path overlap or resource conflict is detected, strategy correction is performed, achieving path avoidance through path replanning, priority adjustment, or waiting mechanisms. Finally, planning instructions are sent to the AGV devices, and running information is transmitted back in real time to update the occupancy status and global planning data. The complete process is as follows: Figure 2 As shown.

[0025] Example 1 like Figure 1 As shown, a multi-AGV collaborative path planning method for a wheel rim production workshop includes the following steps: S1: Model the environment of the wheel rim production workshop to obtain a grid map, corridor topology set, and narrow resource segment set; S2: Map the planned path of the AGV to the corridor topology set, and calculate and manage the occupancy status of each corridor segment in the corridor topology set on the time axis based on the driving status of the AGV, and generate an occupancy status table. S3: Based on the occupancy status table, the improved TRTD-RRT algorithm is used for path planning. When expanding and reconnecting path nodes, the improved TRTD-RRT algorithm takes the occupancy risk factor of the target corridor segment as part of the comprehensive cost function. S4: Perform spatial and temporal conflict detection on each generated AGV path; S5: If a strong conflict is detected, the collaborative scheduling strategy will be activated to adjust the passage priority of the AGV or perform local path replanning according to the preset rules. S6: Convert the final determined path into motion control commands for the AGV and perform closed-loop control based on sensor feedback.

[0026] As a refinement of the above embodiments, step S1 specifically includes: First, environmental data of the workshop is collected using LiDAR. Then, a grid map of passable areas is generated using a gridding method. Finally, the edges of the equipment are expanded using an obstacle expansion algorithm to ensure the safe operating distance of the AGV. The expansion radius formula is as follows.

[0027] , in, Let be the minimum envelope radius of the AGV. This represents the maximum estimated deviation of the positioning and control errors. To ensure the required safe distance for workshop equipment, the expanded map is denoted as M, where a grid value of 1 represents an impassable area and a grid value of 0 represents a passable area. The workshop corridor structure is automatically extracted through corridor edge detection and path connectivity analysis. Finally, the set of all passable segments that satisfy the width, length, and connectivity constraints is defined as the passable segment set C. This passable segment structure serves as the carrier for path constraints and occupancy management. Each corridor segment... Including the starting point ,end and width parameter : , , To identify potential resource conflict risks in the operation of multiple AGVs, based on a width threshold Label the set R of narrow channel segments, where Minimum passable width limit for AGVs: , Finally, the output is structured data obtained through SLAM calculation and topology analysis, including the processed grid map M, corridor topology set C, and narrow resource segment set R, which are used to correct the occupancy risk factor model in the path planning process through width constraints.

[0028] As a refinement of the above embodiments, step S2 specifically includes: Based on the mapping relationship between the path trajectory and the workshop corridor topology, the navigation path is first mapped to the corridor segment set C. When the AGV's movement trajectory passes through a certain corridor segment... At that time, the time interval during which the corridor segment is occupied is calculated based on the path length and vehicle speed: , in, This represents the distance the AGV travels within a certain corridor section. For vehicle operating speed, This represents the time interval for the occupancy of this corridor segment.

[0029] The absolute time range of the occupied interval can be expressed as: , , in, The time point when the AGV enters. This refers to the time of departure. To achieve dynamic occupancy tracking, an occupancy table structure based on "corridor segment + time range" is constructed. Its formal definition is: , in, This is the vehicle identifier with the number p. The absolute time range of the occupied interval is used to determine whether there is a conflict in corridor segment resources before a new path is written into the occupancy table. After the AGV runs along the planned path and leaves the corresponding corridor area, it is determined whether the corridor release condition is met based on the AGV's real-time pose information. If it is detected that the AGV's body has completely left the corridor area, the corridor occupancy status table is automatically updated, and the corresponding corridor segment is marked as idle for subsequent AGV path planning and scheduling.

[0030] As a refinement of the above embodiment, step S3 is the core algorithm. Using the workshop environment model and passageway occupancy information as input, before path searching, the planned space is first pruned according to the time-space occupancy constraints in the occupancy status table. When a candidate path node is located in a passageway area already occupied by other AGVs within the corresponding time window, the node will be determined as a non-expandable node, thereby avoiding the generation of potential conflicting paths during the path generation stage and achieving pre-processing for conflict avoidance. The process is as follows... Figure 3 As shown.

[0031] Specifically, this algorithm uses the traditional bidirectional expanding RRT algorithm as its basic framework, expanding the search tree from both the starting and ending points to accelerate the algorithm's speed. Addressing the problems of redundant path nodes, tortuous paths, and the tendency to form "Z" or "zigzag" paths in traditional RRT algorithms, the algorithm integrates the risk of passage segment occupancy with a local rerouting mechanism during the search tree expansion process. As soon as a new path node is successfully added to the search tree, the local rerouting optimization process is immediately triggered. S301: This new node and its neighborhood radius The node connectivity relationships within the range are evaluated and adjusted, and the neighborhood radius is determined. Used to control the search range during rerouting, thereby determining the computational cost of rerouting, it is defined as follows: , , in, This represents the current number of sampling nodes. It is a constant factor. The area of ​​the feasible space; It is the dimension of the search space, in a two-dimensional workshop environment. The value is 2.

[0032] S302: When making reconnection decisions, this invention no longer uses only the path geometric length as the optimization objective, but instead constructs a comprehensive cost function that integrates path length, curvature constraints, and the risk of passage segment occupancy, defined as: , in, This represents the total cost from the current node to the parent node. As the parent node, For the current node (the new path node). This represents the cumulative cost from the current node to the parent node. This represents the Euclidean distance from the current node to its parent node. The weighting coefficients, which balance length and smoothness, are set to [0.5, 0.6]. This is the risk cost weight, with a value of [10, 15], used to balance path smoothness and traffic safety. The cost of curvature or angle for connecting branches.

[0033] In this way, the algorithm actively avoids high-conflict-probability areas at the topology level, reduces repeated path corrections and local backtracking, and ensures that the generated path satisfies geometric smoothness while taking into account the safety and feasibility of passage in multi-AGV environments, thereby significantly improving the stability and overall efficiency of path planning in complex workshop environments.

[0034] As a refinement of the above embodiments, in order to improve the efficiency of path search and enhance the stability of the algorithm in complex workshop environments, when selecting new path nodes, the algorithm further introduces a passage section occupancy risk factor in the introduced dynamic target bias sampling mechanism. Based on the occupancy status of the corridor corresponding to the target direction within the planning time window, the path expansion is guided to tend towards unoccupied or lower-risk spatial areas.

[0035] Specifically, calculate the dynamic target bias sampling probability and generate random numbers. The value ranges from [0,1]. If a random number is generated... If the probability is less than the dynamic target bias sampling probability, then target bias sampling is performed, and a new path node is determined by extending the path towards the target point by a fixed step size; otherwise, a new path node is randomly selected in free space.

[0036] The dynamic target bias sampling probability function is: , in, For dynamic target bias sampling probability. To minimize the search probability (ensuring the algorithm has global search capability in the initial stage), we take the value [0.05, 0.15]. To maximize the search probability (preventing excessive greed during the search process), we choose [0.7, 0.9]. For the maximum iteration count, For the current iteration count, To control the growth rate, we choose [1,3] to balance the search and convergence. The occupancy risk factor for the corridor passage segment in the target direction is 0≤ ≤1 is used to characterize the probability of path conflict or resource occupation in the area. The higher the risk, the lower the probability of target bias.

[0037] The calculation formula is: , in, For the narrow resource segment defined above, For width parameter, For width threshold, This is the attenuation sensitivity coefficient, with a value range of [0.5, 2.0]. It controls the rate at which the risk factor decreases as the channel width increases. Increase The cost of travel will decrease exponentially, making travel in wide areas extremely low, especially in narrow resource sections. At that time, the risk factor is directly set to its maximum value of 1.

[0038] When the search tree is far from the target area or the risk of the target direction passage being occupied is high. As the value increases, the system automatically reduces the target bias sampling probability to enhance the exploration capability of the feasible space and avoid premature entry into high-risk areas; when the search tree gradually approaches the target area and the risk of occupying the passage section in the target direction is low, Decreasing the value increases the target bias sampling probability, guiding the search tree to converge quickly towards the target region. By introducing this sampling method, this invention transforms the path search strategy from being driven by a single iteration progress to being driven by a dual-factor synergy of "search progress and occupancy status," effectively avoiding the problem of decreased search efficiency in traditional fixed or single dynamic target bias strategies in narrow corridors, areas where multiple AGVs intersect, and dynamic occupancy environments.

[0039] As a refinement of the above embodiments, step S4 specifically includes: This study identifies potential path interaction risks among multiple AGVs in a workshop corridor, encompassing two judgment methods: spatial conflict detection and temporal conflict detection. By projecting the target path onto the corridor topology set and combining it with an occupancy status table, path overlap detection is achieved, thereby ensuring that multiple AGVs do not interfere with each other or experience opposing congestion during operation.

[0040] In the spatial conflict dimension, the path is first discretized into a series of point sequences, and then mapped to the corridor segment structure set. Assume there are two AGVs on a certain corridor segment, occupying areas respectively. and The following methods are used to determine whether a conflict has occurred: conflict: , No conflict: , Similarly, in the time dimension, if two AGVs occupy a certain corridor segment, the time zone... and Similarly, the following methods can be used to determine whether a conflict has occurred: conflict: , No conflict: , If spatial and temporal conflicts occur simultaneously, meaning the paths overlap and their execution times intersect, it is considered a strong conflict. In this case, conflict information will be output immediately to trigger path reconstruction or time delay strategies. If there is only spatial overlap but no temporal overlap, it is considered an acceptable state. If the path segments do not overlap, there is no risk of conflict.

[0041] As a refinement of the above embodiments, step S5 specifically includes: When a strong conflict is detected, the running paths of the conflicting vehicles are planned and dynamically adjusted collaboratively. By comprehensively analyzing the conflicting path segments and conflicting time intervals, the priority of the conflicting vehicles is determined, and a timing control strategy or path alternative is generated, thereby achieving orderly passage of multiple AGVs.

[0042] First, by receiving conflict data, such as the conflict space, time set, and the AGV number where the conflict occurred, the passage priority of each AGV is calculated according to the following evaluation function: , in, For the first The remaining path length of the AGV from the target point. For the first The estimated planning time for each AGV is as follows: This is a length sensitivity factor with values ​​in the range [0.4, 0.6]. This is a time-sensitive factor with values ​​ranging from [0.5, 1.0]. For the first Priority of each AGV.

[0043] When a strong conflict occurs, vehicles with higher communication priorities have priority to pass, while lower-priority AGVs either delay or trigger a local path replanning strategy. If the conflict is short, the system will delay lower-priority vehicles and suspend vehicle speed control commands to avoid the complex calculations involved in path replanning. If the conflict is long or the conflict area is large, the system will trigger local path replanning, constructing a new extension tree for the conflict section within a local space.

[0044] As a refinement of the above embodiments, step S6 specifically includes: The path planning results are converted into executable motion commands to realize the speed control, direction control and dynamic obstacle avoidance control of AGV, and to complete the path tracking and attitude stabilization process.

[0045] Based on the discrete coordinate sequence of the planned path, the system first extracts local geometric features of the path, including path curvature, rate of change of direction, and segmented arc length information. Then, it combines these with the AGV dynamics model to generate reference values ​​for velocity, angular velocity, and acceleration. For straight segments, the system generates a trapezoidal velocity curve constrained by the segment length and an upper limit of acceleration. The maximum speed is Then speed The definition is as follows: , in, To accelerate to The time period To accelerate to The period after which the speed becomes uniform begins This refers to the deceleration period.

[0046] For curved segments, the upper limit of angular velocity is constrained by the path curvature, and the curvature... and angular velocity It is expressed as follows: , , in, For cumulative arc length, and Divided into path coordinate pairs parameters The first derivative, and These are path coordinate pairs. The second derivative of .

[0047] By integrating LiDAR and IMU feedback data, the system acquires real-time vehicle pose, speed, and acceleration status, and performs dynamic error correction based on local obstacle information. When an obstacle enters the safety threshold range, the controller triggers speed limiting or emergency braking logic, and uses obstacle information to update corridor occupancy status and path priority, forming a closed-loop control structure of planning-execution-scheduling. Finally, motion reference values ​​are converted into drive command outputs, enabling the vehicle to track the planned path online and maintain operational stability. The inputs are the final path sequence and sensor feedback data, and the outputs are speed and angular velocity execution commands, realizing an integrated operation process of path execution, obstacle avoidance response, and attitude control.

[0048] Example 2 A multi-AGV collaborative path planning system for a wheel rim production workshop, the system comprising: The environment modeling module is used to build a workshop environment model and output a workshop map, a corridor topology set, and a narrow resource segment set. This module outputs structured data obtained through SLAM calculation and topology analysis, including a processed raster map M, a corridor topology set C, and a narrow resource segment set R. This data is used to support the path planning module in generating feasible paths and to provide a resource segment structure foundation for the corridor occupancy management module, thereby realizing digital structured modeling of the workshop environment space.

[0049] The corridor occupancy management module maps AGV paths to corridor topology and manages the time occupancy status of each corridor segment. Based on the mapping relationship between the path trajectory and the workshop corridor topology, this module first maps the navigation path output by the path planning module to the corridor segment set. When the AGV's trajectory passes through a corridor segment, the system calculates the time interval during which the corridor segment is occupied based on the path length and the vehicle's speed. This module determines whether corridor segment resources conflict by checking for overlap in time intervals. Simultaneously, this module is also responsible for managing the release of corridor resources. When an AGV runs along the planned path and leaves the corresponding corridor area, the system determines whether it meets the corridor release conditions based on the AGV's real-time pose information. If it detects that the AGV's body has completely left the corridor area, the system automatically updates the corridor occupancy status table, marking the corresponding corridor segment as idle for subsequent AGV path planning and scheduling.

[0050] The path planning module is used to plan paths for AGVs based on occupancy status using the TRTD-RRT algorithm. This module is the core algorithm, taking the workshop environment model and passageway occupancy information as input. Before path searching, it first prunes the planning space according to the time-space occupancy constraints output by the corridor occupancy management module. When a candidate path node is located in a passageway area already occupied by other AGVs within the corresponding time window, the node is determined to be a non-expandable node, thus avoiding the generation of potential conflict paths during the path generation stage and achieving pre-processing for conflict avoidance.

[0051] The conflict detection module is used to detect spatial and temporal conflicts between different AGV paths; it identifies potential path interaction risks among multiple AGVs in the workshop corridor space, covering two types of judgment methods: spatial conflict detection and temporal conflict detection. This module projects the target path generated by the path planning module onto the corridor reference topology and combines it with the running time and spatial information stored in the corridor occupancy management module to detect path overlap, thereby ensuring that multiple AGVs do not interfere with each other or cause congestion in opposite directions during operation.

[0052] The collaborative scheduling module is used to make scheduling decisions for conflicting AGVs when strong conflicts are detected. When the conflict detection module identifies a strong conflict state, it performs collaborative planning and dynamic adjustment of the running paths of conflicting vehicles. By comprehensively analyzing the conflict path segments and conflict time intervals, it determines the priority of conflicting vehicles and generates timing control strategies or path alternatives, thereby achieving orderly passage of multiple AGVs.

[0053] The execution control module is used to convert the planned path into AGV motion commands and realize closed-loop control; it also completes the path tracking and attitude stabilization process. Ultimately, the motion reference values ​​are converted into drive command outputs, enabling the vehicle to track and plan its path online while maintaining operational stability. The module inputs are the final path sequence and sensor feedback data, and the module outputs speed and angular velocity execution commands, realizing an integrated operation process of path execution, obstacle avoidance response, and attitude control.

[0054] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-AGV collaborative path planning method for a wheel rim production workshop, characterized in that, Includes the following steps: Model the environment of the wheel rim production workshop to obtain a grid map, corridor topology set, and narrow resource segment set; The planned path of the AGV is mapped to the corridor topology set, and the occupancy status of each corridor segment in the corridor topology set on the time axis is calculated and managed based on the driving status of the AGV, generating an occupancy status table. Based on the occupancy status table, an improved TRTD-RRT algorithm is used for path planning. When expanding and reconnecting path nodes, the improved TRTD-RRT algorithm takes the occupancy risk factor of the target corridor segment as part of the comprehensive cost function. Perform spatial and temporal conflict detection on each generated AGV path; If a strong conflict is detected, a collaborative scheduling strategy is activated, adjusting the AGV's passage priority or performing local path replanning according to preset rules. The finalized path is converted into motion control commands for the AGV, and closed-loop control is performed based on sensor feedback.

2. The multi-AGV collaborative path planning method for wheel rim production workshops according to claim 1, characterized in that, The modeling of the wheel rim production workshop environment specifically includes: The system collects environmental data from the workshop using sensors, generates a grid map, and inflates the obstacle areas. Extract passage segments that meet preset width, length, and connectivity constraints from the processed map to form a corridor topology set; Based on the preset minimum passage width threshold for AGVs, a set of narrow resource segments is identified and marked from the corridor topology set.

3. The multi-AGV collaborative path planning method for wheel rim production workshops according to claim 1, characterized in that, The occupancy status table is used to record the occupancy status of each corridor segment by different AGVs in different time intervals; Based on the AGV's travel distance and preset speed within the corridor section, its entry and exit times are calculated to determine the occupied time interval.

4. The multi-AGV collaborative path planning method for wheel rim production workshops according to claim 1, characterized in that, The improved TRTD-RRT algorithm employs a bidirectional expanding search tree framework, and its improvements include: When a new path node is added to the search tree, a local rewiring is performed, and the comprehensive cost function is used to compare the path node with its neighborhood radius. The connection relationships of the nodes within the node are adjusted, and nodes with lower overall cost functions are selected for connection; The new path nodes are selected based on a dynamic target bias sampling mechanism that considers the occupancy risk factor of the corridor passage segment in the target direction: Calculate the dynamic target bias sampling probability and generate random numbers. If random number If the probability is less than the dynamic target bias sampling probability, then target bias sampling is performed, and a new path node is determined by extending the path towards the target point by a fixed step size; otherwise, a new path node is randomly selected in free space.

5. The multi-AGV collaborative path planning method for wheel rim production workshops according to claim 1, characterized in that, The comprehensive cost function is used to compare the path node with its neighborhood radius. The connection relationships between nodes within the node are adjusted as follows: After generating new path nodes, use this as the center and the neighborhood radius The search term is defined as the range within which all existing nodes within the region are considered as candidate nodes for evaluation. Calculate the combined cost between the new path node and the node to be determined, and select the node with the lower combined cost as the target node for connection.

6. The multi-AGV collaborative path planning method for wheel rim production workshops according to claim 5, characterized in that, The formula for calculating the dynamic target bias sampling probability is as follows: ， in, For dynamic target bias sampling probability. To minimize the search probability, To maximize the search probability, For the maximum iteration count, For the current iteration count, To control the growth rate, The occupancy risk factor for the corridor segment in the target direction is calculated using the following formula: ， in, For a narrow set of resource segments, For width parameter, For width threshold, This is the attenuation sensitivity coefficient.

7. The multi-AGV collaborative path planning method for wheel rim production workshops according to claim 1, characterized in that, The comprehensive cost function is as follows: ， in, This represents the total cost from the current node to the parent node. As the parent node, For the current node, This represents the cumulative cost from the current node to the parent node. This represents the Euclidean distance from the current node to its parent node. The weighting coefficients are used to balance length and smoothness. As a risk cost weight, The cost of curvature or angle for connecting branches.

8. The multi-AGV collaborative path planning method for wheel rim production workshops according to any one of claims 1-7, characterized in that, The collision detection includes: Spatial conflict detection: Determine whether the projected areas of the planned paths of different AGVs overlap on the corridor section; Time conflict detection: Determine whether there is an overlap in the time intervals during which different AGVs occupy the same corridor segment; If both spatial overlap and temporal intersection exist simultaneously, it is considered a strong conflict.

9. The multi-AGV collaborative path planning method for wheel rim production workshops according to claim 8, characterized in that, The coordinated scheduling strategy includes: The passage priority of each conflicting AGV is calculated based on its remaining path length and estimated planning time, using the following formula: ， in, For the first The remaining path length of the AGV from the target point. For the first The estimated planning time for each AGV is as follows: As a length-sensitive factor, As a time-sensitive factor, For the first Priority of each AGV; If a high-priority AGV has a strong conflict with a low-priority AGV, the high-priority AGV will pass first, while the low-priority AGV will either wait or trigger a local path replanning strategy.

10. A multi-AGV collaborative path planning system for a wheel rim production workshop, characterized in that, The system for implementing the method of any one of claims 1 to 9 comprises: The environment modeling module is used to build a workshop environment model and output a workshop map, a corridor topology set, and a narrow resource segment set. The corridor occupancy management module is used to map AGV paths to corridor topology and manage the time occupancy status of each corridor segment. The path planning module is used to plan the path for AGVs based on the occupancy status and using the TRTD-RRT algorithm. The conflict detection module is used to detect spatial and temporal conflicts between different AGV paths; The collaborative scheduling module is used to make scheduling decisions for conflicting AGVs when strong conflicts are detected. The execution control module is used to convert the planned path into AGV motion commands and realize closed-loop control.

Images