Workshop multi-agv path planning method and device

Abstract

The application provides a workshop multi-AGV path planning method and device, and belongs to the technical field of route optimization. The method comprises the following steps: discretizing a path into a node sequence, constructing a global time window matrix, and generating a path-time window combination. The path-time window combination comprises a conflict-free global path formed by the node sequence and a planned time window corresponding to each node in the conflict-free global path. The global path-time window combination is sent to each AGV. Based on the real-time position and speed of each AGV, the deviation of the actual node occupation time of each AGV from the planned time window is calculated. If the deviation of any AGV does not exceed a first deviation threshold, the AGV is instructed to adjust the driving speed. If the deviation of any AGV is greater than the first deviation threshold, the path-time window combination is updated. The application optimizes the driving continuity and task completion timeliness of the AGV.

Description

Technical Field

[0001] This application relates to the field of route optimization technology, and in particular to a method and apparatus for multi-AGV path planning in a workshop. Background Technology

[0002] With the deepening of intelligent manufacturing and industry, Automated Guided Vehicles (AGVs), as core equipment for workshop logistics automation, have been widely used in material handling, production line delivery, and other processes. In complex workshop environments, the collaborative operation of multiple AGVs can significantly improve logistics efficiency, reduce labor costs, and enhance production flexibility. However, with the increase in the number of AGVs and the complexity of tasks, how to perform efficient and conflict-free path planning for multiple AGVs has become crucial to ensuring the stable operation of the workshop logistics system.

[0003] Existing multi-AGV path planning methods in workshops typically employ centralized or distributed strategies. These involve a central server coordinating the planning or individual AGVs planning autonomously. Regardless of the strategy, they largely rely on path search algorithms such as A* and Dijkstra's algorithm, combined with time windows or priority rules for conflict resolution. Some methods simplify the planning process by pre-setting fixed paths or using static scheduling strategies. However, these methods often struggle to respond in real-time to AGV deviations in dynamically changing workshop environments. When an AGV's actual speed deviates from the plan due to factors such as load, ground conditions, or communication delays, the original path planning results may become invalid, leading to node occupancy conflicts, deadlocks, or increased waiting times. Furthermore, existing methods are insufficient in the joint optimization of path and time, lacking a fine-grained allocation and dynamic adjustment mechanism for global time resources, resulting in decreased overall system efficiency and low resource utilization.

[0004] In conclusion, how to avoid conflicts in multi-AGV systems in dynamic environments and optimize operational efficiency has become an urgent technical problem to be solved. Summary of the Invention

[0005] This application provides a method and apparatus for multi-AGV path planning in a workshop, which addresses the shortcomings of existing technologies in dealing with operational deviations and achieving dynamic collaboration.

[0006] This application provides a method for multi-AGV path planning in a workshop, including the following steps: The path is discretized into a sequence of nodes, a global time window matrix is ​​constructed, and a path-time window combination is generated. The path-time window combination includes a conflict-free global path formed by the node sequence and a planned time window corresponding to each node in the conflict-free global path. Distribute the global path-time window combination to each AGV; Based on the real-time position and speed of each AGV, the deviation between the actual node occupancy time of each AGV and the planned time window is calculated; If the deviation of any AGV does not exceed the first deviation threshold, the AGV is instructed to adjust its travel speed; if the deviation of any AGV is greater than the first deviation threshold, the path-time window combination is updated.

[0007] According to the workshop multi-AGV path planning method provided in this application, the generation of path-time window combination includes: Conflict detection steps: Detect whether there are node conflicts or opposing conflicts among the AGVs; Conflict resolution steps: If there are node conflicts or opposing conflicts, the conflicting AGVs are sorted according to preset priorities, and the low-priority AGVs are shifted within a time window or replanned based on a preset path planning algorithm. Repeat the conflict detection step and the conflict resolution step until there are no conflicts in the path-time window combinations of all AGVs.

[0008] According to the workshop multi-AGV path planning method provided in this application, the path planning algorithm adopts an improved TROA algorithm, which is initialized based on the following steps: Determine the dimension of the solution space, and determine a prime number based on the dimension of the solution space; Based on the prime numbers, generate the optimal point vector; Based on the optimal point vector, a unit space point set is generated and mapped to a raster map. The starting point and ending point are forced to be consistent with the task objective, resulting in different paths from the starting point to the ending point.

[0009] According to the workshop multi-AGV path planning method provided in this application, the improved TROA algorithm performs parameter iterative optimization based on the following steps: In the TROA algorithm, the hunting step size, territory range, ambush probability, and random roaming probability are each established with a nonlinear decay function relative to the current iteration progress, where the current iteration progress is the ratio of the current iteration number to the maximum iteration number. Based on the aforementioned nonlinear decay functions, the hunting stride length, the territory range, the ambush probability, and the random roaming probability are updated respectively.

[0010] According to the workshop multi-AGV path planning method provided in this application, the nonlinear decay function of the hunting step length is as follows: ; in, It is the initial hunting stride length. It is the current iteration number. It is the maximum number of iterations. It is the decay index. yes Hunting stride length in the [number]th generation.

[0011] According to the workshop multi-AGV path planning method provided in this application, the nonlinear decay function of the raid probability is as follows: ; in, It is the initial raid probability. It is the current iteration number. It is the maximum number of iterations. It is the decay index. It is the minimum probability of a surprise attack. It is the first The probability of a surprise attack during a certain era.

[0012] According to the workshop multi-AGV path planning method provided in this application, updating the path-time window combination includes: If the deviation of the AGV does not exceed the second deviation threshold, the planned time window of all subsequent nodes of the AGV is adjusted, and it is checked whether the adjusted path-time window combination causes new conflicts with other AGVs. If no new conflicts are caused, the adjusted path-time window combination is sent to the above AGV. The second deviation threshold is greater than the first deviation threshold. If the deviation of the AGV is greater than the second deviation threshold or the adjusted path-time window combination causes a new conflict with other AGVs, the path-time window combination of the affected AGVs is adjusted based on the conflict resolution step.

[0013] This application also provides a method for multi-AGV path planning in a workshop, including the following steps: The system receives a global path-time window combination sent by the server and runs based on its own path-time window combination. The path-time window combination includes a conflict-free global path formed by a sequence of nodes and a planned time window corresponding to each node in the conflict-free global path. When encountering obstacles, an incremental detour path is generated based on local environmental information; based on the estimated length of the detour path, the current travel speed, and the global path-time window combination, it is determined whether the detour path will cause new conflicts with other AGVs. If there is no conflict, the detour is carried out based on the detour path, and the updated path-time window combination is synchronized to the server. If there is a conflict, the local adjustment conflict event and obstacle information are reported to the server.

[0014] This application also provides a workshop multi-AGV path planning device, including the following modules: The global planning module is used to: discretize the path into a sequence of nodes, construct a global time window matrix, and generate a path-time window combination, wherein the path-time window combination includes a conflict-free global path formed by the node sequence and a planned time window corresponding to each node in the conflict-free global path; The path distribution module is used to distribute the global path-time window combination to each AGV. The real-time monitoring module is used to: calculate the deviation between the actual node occupancy time of each AGV and the planned time window based on the real-time position and speed of each AGV; The dynamic adjustment module is used to: instruct the AGV to adjust its travel speed if the deviation of any AGV does not exceed a first deviation threshold; and update the path-time window combination if the deviation of any AGV exceeds the first deviation threshold.

[0015] This application also provides a workshop multi-AGV path planning device, including the following modules: The control module is used to: receive a global path-time window combination sent by the server, and run based on its own path-time window combination, wherein the path-time window combination includes a conflict-free global path formed by a sequence of nodes, and a planned time window corresponding to each node in the conflict-free global path. The real-time detour module is used to: generate an incremental detour path based on local environmental information when encountering obstacles; and determine whether the detour path will cause new conflicts with other AGVs based on the estimated length of the detour path, the current driving speed, and the global path-time window combination. The local obstacle avoidance module is used to: if there is no conflict, detour based on the detour path and synchronize the updated path-time window combination to the server; if there is a conflict, report the local adjustment conflict event and obstacle information to the server.

[0016] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the workshop multi-AGV path planning method as described above.

[0017] This application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the workshop multi-AGV path planning method as described above.

[0018] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the workshop multi-AGV path planning method as described above.

[0019] The workshop multi-AGV path planning method and apparatus provided in this application discretizes the path into a sequence of nodes and constructs a global time window matrix to generate a path-time window combination containing a conflict-free global path and the planned time windows of each node, thereby achieving joint optimization of the running path and time of multiple AGVs. Furthermore, based on the real-time position and speed of each AGV, the deviation between its actual node occupancy time and the planned time window is calculated. When the deviation does not exceed a first deviation threshold, only the AGV is instructed to adjust its travel speed, without replanning the path, thus correcting the running deviation at minimal cost and maintaining system stability. When the deviation exceeds the first deviation threshold, the path-time window combination is updated, and global coordination is re-performed to avoid conflicts or deadlocks caused by the accumulation of local deviations. This application achieves refined allocation of AGV spatiotemporal resources through the time window mechanism, effectively avoiding conflicts between multiple AGVs at nodes and on paths, improving the system's adaptability to dynamic disturbances, and reducing the computational overhead caused by frequent replanning since path adjustments are only performed when necessary, thus improving planning efficiency and system real-time performance. In summary, this application optimizes the continuity of AGV operation and the timeliness of task completion while ensuring conflict-free operation, thereby improving the overall operating efficiency and resource utilization of the workshop logistics system. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is one of the flowcharts illustrating the workshop multi-AGV path planning method provided in this application; Figure 2 This is a flowchart of the multi-AGV conflict detection and resolution based on the improved dynamic time window provided in this application; Figure 3 This application provides the TROA algorithm for initializing population distribution maps based on random initialization and optimal point set initialization, respectively. Figure 4 This is a flowchart of the global path planning process for AGVs based on the improved TROA algorithm provided in this application; Figure 5 This is the second flowchart of the workshop multi-AGV path planning method provided in this application; Figure 6 This is the third flowchart of the workshop multi-AGV path planning method provided in this application; Figure 7 This is a system framework diagram of the workshop multi-AGV path planning method provided in this application; Figure 8 This is one of the structural schematic diagrams of the workshop multi-AGV path planning device provided in this application; Figure 9 This is the second structural schematic diagram of the workshop multi-AGV path planning device provided in this application; Figure 10 This is a schematic diagram of the structure of the electronic device provided in this application. Detailed Implementation

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

[0023] It should be noted that in the description of the embodiments of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. The terms "upper," "lower," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application. Unless otherwise expressly specified and limited, the terms "installed," "connected," and "linked" should be interpreted broadly, for example, they can be fixed connections, detachable connections, or integral connections; they can be mechanical connections or electrical connections; they can be direct connections or indirect connections through an intermediate medium; and they can be internal connections between two elements. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0024] The terms "first," "second," etc., used in this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such numbers can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, a first object can be one or more. Furthermore, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects have an "or" relationship.

[0025] The following is combined Figures 1-10 This application describes the workshop multi-AGV path planning method and apparatus provided in the embodiments.

[0026] Figure 1 This is one of the flowcharts illustrating the workshop multi-AGV path planning method provided in this application, such as... Figure 1 As shown, the method includes the following: S110, Discretize the path into a node sequence, construct a global time window matrix, and generate a path-time window combination. The path-time window combination includes a conflict-free global path formed by the node sequence and a planned time window corresponding to each node in the conflict-free global path. S120 distributes the global path-time window combination to each AGV; S130, Based on the real-time position and speed of each AGV, calculate the deviation between the actual node occupancy time of each AGV and the planned time window; S140, if the deviation of any AGV does not exceed the first deviation threshold, the AGV is instructed to adjust its travel speed; if the deviation of any AGV is greater than the first deviation threshold, the path-time window combination is updated.

[0027] It should be noted that the execution subject of the workshop multi-AGV path planning method provided in this application embodiment is the server side. Specifically, it can be a server, computer equipment, such as mobile phone, tablet computer, laptop computer, handheld computer, vehicle electronic equipment, wearable device, ultra-mobile personal computer (UMPC), netbook or personal digital assistant (PDA), etc.

[0028] For ease of understanding, the following embodiments use a workshop server as the execution subject to illustrate the workshop multi-AGV path planning method of the server side of this application.

[0029] In S110, the node sequence is a set of points discretized and topologically represented by the continuous physical path traveled by the AGV. The global time window matrix is ​​a multi-dimensional spatiotemporal resource allocation data structure. The rows and columns of the matrix can represent different AGVs and path nodes, respectively. The elements in the matrix record the planned time window that each AGV is allowed to occupy at a specific node.

[0030] In this embodiment, the workshop server obtains the current task start and end points of all AGVs, plans a conflict-free path in the topology map, converts it into a node sequence, calculates the initial time window of each node, constructs a global time window matrix, and allocates a planned time window to each node based on the rated speed and turning time of each AGV, thereby obtaining a conflict-free global path-time window combination.

[0031] In the specific implementation process, the workshop floor plan is divided into square grids of fixed size, with the center point of each grid serving as a node.

[0032] In S120, the generated path-time window combination is packaged and sent to the controller of each AGV via wireless network.

[0033] In S130, the AGV acquires its position and speed in real time through onboard sensors during its journey and reports them to the workshop server at high frequency. After receiving the data, the workshop server calculates the actual node occupancy time for the AGV to reach the next or subsequent node based on the AGV's current actual position and speed. The calculated actual node occupancy time is compared with the corresponding planned time window in the global time window matrix to calculate the time deviation. For example, if the planned arrival time is 10:00:00 and the actual estimated arrival time is 10:00:05, the deviation is +5 seconds.

[0034] In S140, the calculated absolute value of the deviation is compared with the first deviation threshold. If the deviation is within the allowable range, it indicates that the system is in a normal fluctuation state. The server only issues a speed adjustment command to the AGV, instructing the AGV to accelerate and catch up, without changing its original path and subsequent time window nodes. If the deviation exceeds the first deviation threshold, the original time window system faces the risk of collapse. The server will trigger a replanning mechanism, using the current actual position of the AGV as the starting point, to recalculate and update the global path-time window combination, and then issue it to the affected AGV.

[0035] Here, the first deviation threshold can be adaptively set according to actual usage requirements.

[0036] The workshop multi-AGV path planning method provided in this application discretizes the path into a sequence of nodes and constructs a global time window matrix to generate a path-time window combination containing a conflict-free global path and the planned time windows of each node. This achieves joint optimization of the running path and time of multiple AGVs. Furthermore, based on the real-time position and speed of each AGV, the deviation between its actual node occupancy time and the planned time window is calculated. When the deviation does not exceed a first deviation threshold, only the AGV is instructed to adjust its speed; there is no need to replan the path, thus correcting the running deviation at minimal cost and maintaining system stability. When the deviation exceeds the first deviation threshold, the path-time window combination is updated, and global coordination is re-performed to avoid conflicts or deadlocks caused by the accumulation of local deviations. This application achieves refined allocation of AGV spatiotemporal resources through the time window mechanism, effectively avoiding conflicts between multiple AGVs at nodes and on paths, improving the system's adaptability to dynamic disturbances, and reducing the computational overhead caused by frequent replanning since path adjustments are only performed when necessary, thus improving planning efficiency and system real-time performance. In summary, this application optimizes the continuity of AGV operation and the timeliness of task completion while ensuring conflict-free operation, thereby improving the overall operating efficiency and resource utilization of the workshop logistics system.

[0037] In an optional embodiment, the generation path-time window combination includes: Conflict detection steps: Detect whether there are node conflicts or opposing conflicts among the AGVs; Conflict resolution steps: If there are node conflicts or opposing conflicts, the conflicting AGVs are sorted according to preset priorities, and the low-priority AGVs are shifted within a time window or replanned based on a preset path planning algorithm. Repeat the conflict detection step and the conflict resolution step until there are no conflicts in the path-time window combinations of all AGVs.

[0038] Here, node conflict refers to two or more AGVs attempting to occupy the same node within the same planned time window; opposing conflict refers to two AGVs traveling in opposite directions on the same two-way or one-way prohibited path, and their planned time windows overlap.

[0039] Here, the preset priority is a predefined AGV scheduling weight. The priority can be set directly based on the task assignment order, or it can be set comprehensively based on the urgency of the task, the current load status of the AGV, the distance from the target point, or the type of the AGV itself (such as forklift AGVs having a higher priority than ordinary handling AGVs).

[0040] Here, time window shifting involves artificially increasing the waiting time or adjusting the travel speed of the AGV without changing its spatial travel path, thereby shifting the planned occupancy time of the AGV at the conflict node backward and avoiding overlap with the time window of a high-priority AGV.

[0041] like Figure 2 As shown, the workshop server discretizes the global path of each AGV into a sequence of nodes, and calculates the estimated entry time of each node based on the AGV kinematic model. / Estimated departure time The initial node time windows of each AGV are obtained. A global time window matrix is ​​constructed based on the node time windows of all AGVs. The existence of node conflicts or opposing conflicts is detected. If a conflict exists, the conflicting AGVs are sorted according to a preset priority. Time window shifting or path replanning is performed on low-priority AGVs to resolve the conflict, and the time window matrix is ​​updated. This step is repeated until there are no conflicts in the path-time window combinations of all AGVs.

[0042] In the specific implementation process, during the conflict resolution step, the workshop server extracts the AGVs involved in the conflict and sorts them according to the preset priority rules. High-priority AGVs retain their original paths and time windows. For low-priority AGVs, the system first attempts to shift the time window, calculates the amount of time that needs to be delayed or advanced, and updates the time windows of their subsequent nodes. If the time window shift cannot resolve the conflict, for example, if the shift causes new conflicts with other AGVs, or if the shift time exceeds the maximum delay allowed by the task, then path replanning is triggered, and the path is replanned based on the preset path planning algorithm.

[0043] The workshop multi-AGV path planning method provided in this application, through a conflict detection step, can accurately locate node conflicts and oncoming conflicts. Through a mandatory conflict resolution step, it ensures that within the same spatiotemporal dimension, any physical node or path segment is uniquely and exclusively occupied, eliminating the possibility of collision hazards in issued instructions and guaranteeing the physical safety of the workshop. Furthermore, by prioritizing tasks, core tasks are given priority in obtaining right-of-way, avoiding a decrease in overall efficiency caused by secondary tasks blocking critical passages.

[0044] In an optional embodiment, the path planning algorithm employs a modified TROA algorithm, which is initialized based on the following steps: Determine the dimension of the solution space, and determine a prime number based on the dimension of the solution space; Based on the prime numbers, generate the optimal point vector; Based on the optimal point vector, a unit space point set is generated and mapped to a raster map. The starting point and ending point are forced to be consistent with the task objective, resulting in different paths from the starting point to the ending point.

[0045] Here, TROA (Tyrannosaurus Rex Optimization Algorithm) is a novel swarm intelligence optimization algorithm inspired by the hunting behavior of Tyrannosaurus Rex. While the standard version of this algorithm aims to balance global exploration and local exploitation, its actual optimization capabilities have certain limitations, such as insufficient global search and a tendency to get trapped in local optima. Therefore, in engineering applications, it is often necessary to combine it with improvement strategies to enhance performance. In algorithm optimization, the solution space dimension represents the number of variables required to solve the problem. In path planning scenarios, the solution space dimension typically corresponds to the number of nodes that need to be determined in the path sequence.

[0046] Here, the optimal point vector is a vector generated based on the optimal point set theory in number theory. The optimal point set is a uniform sampling method in high-dimensional space that can cover the entire solution space more evenly with fewer sample points and avoid sample clustering.

[0047] Here, the unit space point set refers to the set of multidimensional coordinate points generated in the normalized interval [0, 1]. These points represent the relative positions of the algorithm initialization and need to be converted into actual physical coordinates on the workshop grid map through a mapping function.

[0048] In this embodiment, the workshop server determines the number of nodes to be searched for this path planning based on the current AGV's task start and end points and the complexity of the workshop map, i.e., the solution space dimension D. Based on this dimension, a specific prime number is determined, such as selecting the smallest prime number greater than D, or calculating it according to an empirical formula. Using the selected prime number, a high-dimensional optimal point vector is generated through a optimal point set generation formula, such as a sequence of irrational numbers based on the square roots of prime numbers. This vector has excellent uniform distribution characteristics in multi-dimensional space and can represent the high-quality distribution of the initial population. The generated optimal point vector is normalized to obtain a unit space point set. Through a coordinate transformation function, the unit space point set is mapped onto the actual grid map of the workshop, forming multiple initial candidate paths. During the mapping process, a forced alignment operation is performed. Regardless of the initial value of the optimal point vector, the algorithm will forcibly replace the first node of each candidate path with the real start point and the last node with the real end point, ensuring the consistency of all initial paths in terms of the task objective.

[0049] In some embodiments, a dimension-prime number lookup table is pre-established. For example, when the solution space dimension is between 10 and 20, the prime number 29 is fixed; when the dimension is between 20 and 50, the prime number 53 is selected. This method is extremely fast and suitable for scenarios with high real-time requirements.

[0050] In some embodiments, the algorithm calculates in real time, for example, by multiplying the spatial dimension D by the golden ratio and rounding it up, and then finding the smallest prime number greater than that value as p. This approach is better suited to tasks of different complexities.

[0051] In one specific embodiment, the AGV path is represented as a two-dimensional coordinate sequence of a series of key path points. Assume the path contains... D Key points (dimensions), each individual It consists of two-dimensional coordinates: ; The starting point and the end point The starting point S and ending point E are fixed as given by the task, with the middle point as follows: D- Two points can move freely within the grid area, all coordinates are rounded to the nearest grid center, and it is ensured that adjacent points are reachable within 8 neighborhoods and that the path does not cross obstacles.

[0052] Based on this, a set of optimal points is used to initialize the population, ensuring that individuals are uniformly covered in the solution space. Population initialization refers to generating different AGV paths from the start point to the end point. The steps for generating the set of optimal points include: Determine the dimension and prime number: Let the dimension of the solution space be . (Each path point contains x, y coordinates), take prime numbers. satisfy ,generally Find the smallest prime number that satisfies the condition; Constructing optimal point vectors: Generating optimal point vectors Each component is composed of It is confirmed that the formula is derived from the projection of equally divided points on the unit circle, and because... Being prime numbers, they possess good independence and uniformity; Generate a unit spatial point set and map it to the actual raster map: For the first... individual In its first dimensional coordinate components Calculate using the following formula: ; in and The first The lower and upper bounds of the dimensional coordinates correspond to the row and column ranges of the raster map. Continuous coordinates are rounded to the nearest raster center coordinates, and the start and end points are forced to be consistent with the given values ​​in the task.

[0053] By initializing the population through the above steps, different initial AGV paths are obtained, providing a high-quality and highly diverse initial population for improving the TROA algorithm. This allows the algorithm to have a broad global search perspective from the initial run, effectively avoiding premature convergence and laying a solid foundation for finding the globally optimal AGV cooperative path in the future.

[0054] Furthermore, to verify the effectiveness of the optimal point set initialization method, a population initialization comparison experiment was conducted on a 15×15 raster map. The starting point was fixed at (1,1), the ending point was fixed at (14,14), and the population size was set to 80 individuals. Each individual's path consisted of the starting point, an intermediate point, and the ending point. In the random initialization method, the coordinates of the intermediate points were uniformly and randomly sampled within the range of 0 to 14. The optimal point set initialization method used the above steps to generate optimal point vectors, which were then mapped to raster coordinates and rounded to the nearest integer. Figure 3 The results of population initialization are presented. It can be seen that the intermediate points and paths generated by random initialization exhibit random clusters and blank areas in space, with some areas showing dense overlap and others lacking coverage, resulting in poor distribution uniformity. In contrast, the intermediate points and paths generated by optimal point set initialization achieve uniform coverage throughout the entire grid space, with no obvious clusters or blank areas. The paths from the start point to the end point are evenly distributed, demonstrating theoretically low bias characteristics. This comparison shows that optimal point set initialization, through deterministic low-bias sequences, achieves more uniform coverage in the solution space than random initialization. This characteristic helps improve the TROA algorithm, enabling higher population diversity and a wider search horizon in the initial stage, fundamentally reducing the risk of premature convergence and laying a high-quality starting point for subsequent global path optimization.

[0055] The workshop multi-AGV path planning method provided in this application uses the theory of optimal point sets in number theory to generate optimal point vectors based on prime numbers. According to mathematical theory, the deviation of optimal point sets in high-dimensional space is extremely small. This means that the generated unit space point set can fill the entire solution space most evenly with the fewest number of samples. Therefore, the algorithm can obtain a highly representative and diverse initial path candidate set at the start stage.

[0056] In an optional embodiment, the improved TROA algorithm performs parameter iterative optimization based on the following steps: In the TROA algorithm, the hunting step size, territory range, ambush probability, and random roaming probability are each established with a nonlinear decay function relative to the current iteration progress, where the current iteration progress is the ratio of the current iteration number to the maximum iteration number. Based on the aforementioned nonlinear decay functions, the hunting stride length, the territory range, the ambush probability, and the random roaming probability are updated respectively.

[0057] Here, the hunting stride is the step size coefficient by which the Tyrannosaurus Rex moves towards prey or the global optimum during the prey search phase of the algorithm. The stride determines the search range of the algorithm in the global space. The territory range is the range by which the Tyrannosaurus Rex marks its territory to avoid competition with other Tyrannosaurus Rex during the territory defense phase of the algorithm. In the algorithm, this corresponds to the diversity maintenance mechanism. If the range is too large, it will easily lead to search divergence; if it is too small, it will easily get trapped in local optima. The ambush probability is the probability that the Tyrannosaurus Rex will launch a fatal ambush after tracking prey. In the algorithm, this corresponds to the mechanism of escaping local optima, which is usually achieved through probabilistic large stride jumps. The random roaming probability is the probability that the Tyrannosaurus Rex will randomly patrol within its territory. In the algorithm, this corresponds to the mechanism of exploring unknown areas or re-initializing its position, which is used to maintain population diversity.

[0058] In this embodiment, at the start of each iteration, the server obtains the current iteration count and the preset maximum iteration count, and uses the ratio of the two as the current iteration progress. The system has a built-in nonlinear decay function model for four core parameters, such as an exponential decay model or a power function model. The current iteration progress is substituted into these four functions to calculate the dynamic values ​​of the four parameters in the current iteration cycle. As the iteration progress increases, the hunting step size and territory range exhibit nonlinear decay. In the early stages of iteration, the iteration progress is small, the parameter values ​​are large, and the algorithm performs large-scale random roaming and global search. In the later stages of iteration, the iteration progress approaches 1, the parameter values ​​decrease sharply, and the algorithm switches to fine-grained local exploration with small steps. The ambush probability and random roaming probability are also dynamically updated based on the nonlinear decay function. In the early stages, a high ambush and roaming probability is maintained to enhance the population's ability to escape local optima and maintain diversity. In the later stages, as the nonlinearity of the probability decreases, the algorithm gradually converges to the vicinity of the global optimum. Using the updated four parameters, the Tyrannosaurus Rex population is driven to execute the position update formula to complete the current iteration.

[0059] In the specific implementation process, let the current iteration number be . The maximum number of iterations is Define the following parameter variables: : No. Hunting step length in a generation, which is the number of grid squares a waypoint moves in a single move; : No. The territory of the time, that is, the Manhattan radius (number of grids) of the local perturbation search. : No. The probability of a surprise attack during a generation; : No. The probability of random roaming during a generation.

[0060] The initial values ​​of the above parameters ( The values ​​of time are denoted as follows: , , , These settings can be configured within the following preferred ranges based on the actual size of the workshop grid map, the motion characteristics of the AGV, and the task requirements: (Number of grid cells) (Number of grid cells) , .

[0061] To ensure the algorithm retains necessary randomness in its later stages to avoid complete stagnation, minimum values ​​are set for the attack probability and random roaming probability, with their preferred ranges being as follows: and In addition, a decay index is introduced. , , , This is used to control the nonlinearity of the decay of each parameter with iteration. The change of parameter values ​​with the number of iterations is determined by the following formula: ; ; ; ; In the above formula, , , , , , , , , , All parameters are configurable. In practical applications, they can be determined through a small number of experiments based on the specific workshop environment, or empirical values ​​can be directly adopted, for example, taking... , , , , , , Good performance can be obtained.

[0062] By employing a parameter adaptive decay mechanism, it is possible to achieve decay in the early stages of iteration. The algorithm employs large step sizes, wide range, and high-probability random behavior for global exploration, rapidly searching a broad region in the solution space, thus effectively avoiding getting trapped in local optima; in the later stages of iteration... The algorithm focuses on fine-grained local exploration near the current optimal solution, improving convergence accuracy and path smoothness. This allows the algorithm to automatically adjust the intensity of exploration and development throughout the optimization process without manual intervention, enhancing its adaptability to workshop maps of varying complexity and size. The final flowchart of the AGV global path planning based on the improved TROA algorithm is shown below. Figure 4 As shown.

[0063] The workshop multi-AGV path planning method provided in this application uses a nonlinear decay function to maintain these parameters at a high level in the early stage of iteration, giving the algorithm global exploration capability. As the iteration progresses, the nonlinear function makes the parameters transition smoothly and gradually decrease, guiding the algorithm to naturally transition from large-scale roaming to small-scale fine search, thus achieving a dynamic balance between global exploration and local development.

[0064] In an optional embodiment, updating the path-time window combination includes: If the deviation of the AGV does not exceed the second deviation threshold, the planned time window of all subsequent nodes of the AGV is adjusted, and it is checked whether the adjusted path-time window combination causes new conflicts with other AGVs. If no new conflicts are caused, the adjusted path-time window combination is sent to the above AGV. The second deviation threshold is greater than the first deviation threshold. If the deviation of the AGV is greater than the second deviation threshold or the adjusted path-time window combination causes a new conflict with other AGVs, the path-time window combination of the affected AGVs is adjusted based on the conflict resolution step.

[0065] Traditional time window methods resolve conflicts once offline, and the time window remains fixed. However, in the dynamic environment of a workshop, the actual operating speed of AGVs inevitably fluctuates, or they may encounter obstacles and need to detour, causing the actual node occupancy time to deviate from the planned time window. Traditional methods simply require the AGV to wait in place until the time window aligns, resulting in low efficiency and a high risk of cascading delays. To address these issues, the improved dynamic time window method proposed in this application, based on the aforementioned execution framework, makes core improvements to the traditional time window method, enabling it to possess real-time sensing and dynamic adjustment capabilities.

[0066] Specifically, when the AGV's deviation exceeds the first deviation threshold, the server first determines whether the deviation exceeds the second deviation threshold. If the deviation does not exceed the second deviation threshold, indicating a moderate delay, the server takes a lightweight intervention: only adjusting the planned time windows of all subsequent nodes of the AGV, for example, by extending the overall delay time while keeping its spatial path unchanged. The workshop server simulates the adjusted new time window in memory and checks whether the AGV's new time window will cause new conflicts with the existing time windows of other AGVs in the workshop. If there are no new conflicts, it means that the AGV's delay is within the system's current buffer capacity, and the workshop server directly sends the adjusted path-time window combination to the AGV, which continues to execute its task according to the new schedule. If the deviation exceeds the second deviation threshold, or if the deviation does not exceed the second threshold but the time window shift causes new conflicts, it means that local adjustments cannot solve the problem. In this case, the workshop server triggers global intervention, calling the aforementioned conflict resolution steps to re-coordinate and adjust the path-time window combinations of all affected AGVs, completely eliminating spatiotemporal overlap.

[0067] Here, the second deviation threshold can be adaptively set according to actual usage requirements.

[0068] In one embodiment, the workshop server receives the real-time position and speed of all AGVs at high frequency, calculates the actual entry / exit time of each AGV at the node, and compares it with the planned time window to obtain the deviation. ; like (e.g., 2 seconds): This is considered a minor deviation, which the AGV can recover by fine-tuning its travel speed without intervention; like (e.g., 2-5 seconds): The workshop server actively adjusts the time window of all subsequent nodes of this AGV (overall translation) and quickly checks whether there are any new conflicts with other AGVs; if there are no new conflicts, the adjusted time window is issued for execution; like If a new conflict arises after adjustment (e.g., 5 seconds), a local rescheduling is triggered. Based on factors such as task priority, the affected AGVs are reassigned to a time window or replanned for a path (by calling the improved TROA algorithm).

[0069] The workshop multi-AGV path planning method provided in this application prioritizes adjusting only the time window when the deviation is between the first and second thresholds. Since the time window adjustment only involves simple numerical addition, subtraction or deduction, the computational complexity is extremely low, far lower than the complexity of the path replanning algorithm. Only when such lightweight adjustment cannot maintain system safety or the delay is extremely serious will the global conflict resolution mechanism be used. This hierarchical strategy greatly saves computational resources.

[0070] Figure 5This is the second flowchart of the workshop multi-AGV path planning method provided in this application, as shown below. Figure 5 As shown, the method includes the following: S510, receive the global path-time window combination sent by the server, and run based on its own path-time window combination. The path-time window combination includes a conflict-free global path formed by a sequence of nodes, and the planned time window corresponding to each node in the conflict-free global path. S520: When encountering obstacles, an incremental detour path is generated based on local environmental information; based on the estimated length of the detour path, the current driving speed, and the global path-time window combination, it is determined whether the detour path will cause new conflicts with other AGVs. S530, if there is no conflict, detour based on the detour path and synchronize the updated path-time window combination to the server; if there is a conflict, report the local adjustment conflict event and obstacle information to the server.

[0071] It should be noted that the execution subject of the workshop multi-AGV path planning method provided in this application embodiment is the AGV end, specifically, the controller of the AGV end.

[0072] In this embodiment, the AGV's onboard controller receives a global path-time window combination from the server via a wireless network (such as 5G / Wi-Fi). The AGV parses this combination to obtain its own conflict-free global path node sequence and the planned time window for each node, and performs autonomous navigation and driving strictly according to this plan. During driving, the AGV perceives the local environment in real time through onboard sensors (such as LiDAR, depth camera, ultrasonic sensors, etc.). When it detects a sudden obstacle not marked on the global map (such as temporarily stacked goods, personnel, etc.), the AGV triggers the local path planning module to generate an incremental detour path based on the local environment information. The AGV extracts the estimated length of this detour path and combines it with the current driving speed. The AGV calculates the time required to execute the detour path. It then compares the calculated detour time window with the global path-time window combination it receives to determine whether the detour will occupy the planned nodes of other AGVs, thus predicting whether new spatiotemporal conflicts will occur with other AGVs. If the prediction result is no conflict, the AGV directly executes the detour path. After the detour is completed, the AGV synchronously sends the updated actual path-time window combination (or detour time data) back to the server so that the workshop server can update the global status. If the prediction result is a conflict, the AGV immediately stops the detour or takes safety braking measures and reports the specific location information of the local adjustment conflict event and obstacles to the server for global overall scheduling.

[0073] In some embodiments, the D* Lite algorithm is used to generate incremental detour paths based on local environmental information.

[0074] The workshop multi-AGV path planning method provided in this application combines local obstacle avoidance decision-making with global conflict prediction, giving the AGV end the ability to autonomously cope with sudden obstacles. Under the premise of ensuring the global conflict-free safety of the multi-AGV system, it greatly reduces the frequent global replanning caused by local sudden situations, significantly reduces the computing pressure and communication latency on the server side, and improves the overall operating efficiency and robustness of the workshop logistics system.

[0075] In summary, the workshop multi-AGV path planning method provided in this application includes two stages, such as... Figure 6 As shown, the first stage is the offline / initial planning stage, which specifically includes: Task reception and global path generation: The workshop server receives a set of multi-AGV tasks (including start point, end point, task priority, and process time constraints) from the upper management system, and calls the improved TROA algorithm adapted to the grid path planning problem to generate an initial global path for each AGV (the path is represented by a sequence of key turning point coordinates). Dynamic time window conflict detection: The workshop server discretizes the global path of each AGV into a sequence of nodes, and calculates the estimated entry time of each node based on the AGV kinematic model. / Estimated departure time The initial node time window for each AGV is obtained. Conflict Detection and Resolution: A global time window matrix is ​​constructed based on the node time windows of all AGVs to detect whether there are node conflicts (multiple vehicles occupying the same node at the same time) or head-on conflicts. If a conflict exists, the AGVs with conflict are sorted according to a preset priority (such as the order in which tasks are assigned). For low-priority AGVs, time window shifting (delaying passage through the conflicting node) or path replanning (using the improved TROA algorithm to generate a new path) is performed to resolve the conflict, and the time window matrix is ​​updated. This step is repeated until there are no conflicts in the path-time window combinations of all AGVs. Path and time window distribution: The workshop server packages and distributes the conflict-free global path and its corresponding node time window constraint table to each AGV.

[0076] The second phase is the online execution phase, which specifically includes: AGV Autonomous Execution and Local Obstacle Avoidance: The AGV travels along the global path while its onboard sensors perceive the environment in real time. Specifically, without dynamic obstacles: it operates strictly according to the global path and time window, and the workshop server continuously monitors the position and time window deviations. Encountering dynamic obstacles: The D* Lite local obstacle avoidance module is triggered, generating an incremental detour path based on local environmental information. Based on the estimated length of the detour path and the current vehicle speed, combined with locally stored global node time window snapshots, it predicts whether the detour will cause new node occupancy conflicts with other AGVs. If there is no conflict, the AGV autonomously executes the detour path, automatically returns to the original global path after detour, and synchronizes the updated position and time window occupancy deviations to the workshop server. If a conflict is predicted, the AGV pauses the detour and reports the local adjustment conflict event and obstacle information to the workshop server, awaiting instructions from the workshop server. If the detour fails (e.g., the passage is completely blocked): The AGV reports a local planning failure event and obstacle information, awaiting instructions from the workshop server. Workshop server real-time collaborative monitoring and anomaly handling: The workshop server receives the real-time position and speed of all AGVs at high frequency, calculates the deviation between the actual node occupancy time and the planned time window, and performs hierarchical dynamic adjustments according to the preset deviation threshold: minor deviations are absorbed by the AGVs themselves by fine-tuning their speed; moderate deviations are handled by the workshop server by shifting the time windows of subsequent nodes as a whole and quickly detecting conflicts; severe deviations or local obstacle avoidance failures trigger local rescheduling, reallocating time windows or replanning paths for the affected AGVs, until a new conflict-free path-time window set is obtained and then sent out for execution, ensuring that the system always maintains a conflict-free operating state in a dynamic environment; The workshop server continues to perform closed-loop monitoring in the second phase until all AGVs have completed their tasks.

[0077] Overall system such as Figure 7 As shown. It should be noted that, Figure 7 A schematic diagram of an exemplary architecture is shown. For descriptive purposes, the architecture depicted is merely an example for a suitable environment and does not impose any limitation on the scope or functionality of this application. Nor should this system be construed as an extension of... Figure 7 Any component shown or a combination thereof has any dependencies or requirements.

[0078] The workshop multi-AGV path planning device provided in the embodiments of this application is described below. The workshop multi-AGV path planning device described below and the workshop multi-AGV path planning method described above can be referred to in correspondence.

[0079] Figure 8 This is one of the structural schematic diagrams of the workshop multi-AGV path planning device provided in this application, such as... Figure 8 As shown, the multi-AGV path planning device in this workshop may include, but is not limited to: The global planning module 810 is used to: discretize the path into a sequence of nodes, construct a global time window matrix, and generate a path-time window combination, wherein the path-time window combination includes a conflict-free global path formed by the node sequence and a planned time window corresponding to each node in the conflict-free global path. The path distribution module 820 is used to distribute the global path-time window combination to each AGV. The real-time monitoring module 830 is used to: calculate the deviation between the actual node occupancy time of each AGV and the planned time window based on the real-time position and speed of each AGV; The dynamic adjustment module 840 is used to: instruct the AGV to adjust its travel speed if the deviation of any of the AGVs does not exceed a first deviation threshold; and update the path-time window combination if the deviation of any of the AGVs exceeds the first deviation threshold.

[0080] It should be noted that the workshop multi-AGV path planning device provided in this application embodiment can execute the workshop multi-AGV path planning method described in any of the above server-side embodiments during actual operation, which will not be elaborated in this embodiment.

[0081] Figure 9 This is the second structural schematic diagram of the workshop multi-AGV path planning device provided in this application, as shown below. Figure 9 As shown, the multi-AGV path planning device in this workshop may include, but is not limited to: The control module 910 is used to: receive a global path-time window combination sent by the server, and run based on its own path-time window combination, wherein the path-time window combination includes a conflict-free global path formed by a sequence of nodes, and a planned time window corresponding to each node in the conflict-free global path. The real-time detour module 920 is used to: generate an incremental detour path based on local environmental information when encountering obstacles; and determine whether the detour path will cause new conflicts with other AGVs based on the estimated length of the detour path, the current driving speed, and the global path-time window combination. The local obstacle avoidance module 930 is used to: if there is no conflict, detour based on the detour path and synchronize the updated path-time window combination to the server; if there is a conflict, report the local adjustment conflict event and obstacle information to the server.

[0082] It should be noted that the workshop multi-AGV path planning device provided in this application embodiment can execute the workshop multi-AGV path planning method described in any of the above AGV end embodiments during specific operation, which will not be elaborated in this embodiment.

[0083] Figure 10An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 10 As shown, the electronic device may include: a processor 1010, a communications interface 1020, a memory 1030, and a communication bus 1040, wherein the processor 1010, the communications interface 1020, and the memory 1030 communicate with each other via the communication bus 1040. The processor 1010 can call logical instructions in the memory 1030 to execute a workshop multi-AGV path planning method, which includes: The path is discretized into a sequence of nodes, a global time window matrix is ​​constructed, and a path-time window combination is generated. The path-time window combination includes a conflict-free global path formed by the node sequence and a planned time window corresponding to each node in the conflict-free global path. Distribute the global path-time window combination to each AGV; Based on the real-time position and speed of each AGV, the deviation between the actual node occupancy time of each AGV and the planned time window is calculated; If the deviation of any of the AGVs does not exceed a first deviation threshold, the AGV is instructed to adjust its travel speed; if the deviation of any of the AGVs exceeds the first deviation threshold, the path-time window combination is updated; or, it includes: The system receives a global path-time window combination sent by the server and runs based on its own path-time window combination. The path-time window combination includes a conflict-free global path formed by a sequence of nodes and a planned time window corresponding to each node in the conflict-free global path. When encountering obstacles, an incremental detour path is generated based on local environmental information; based on the estimated length of the detour path, the current travel speed, and the global path-time window combination, it is determined whether the detour path will cause new conflicts with other AGVs. If there is no conflict, the detour is carried out based on the detour path, and the updated path-time window combination is synchronized to the server. If there is a conflict, the local adjustment conflict event and obstacle information are reported to the server.

[0084] Furthermore, the logical instructions in the aforementioned memory 1030 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0085] On the other hand, this application also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the workshop multi-AGV path planning method provided by the above methods, the method including: The path is discretized into a sequence of nodes, a global time window matrix is ​​constructed, and a path-time window combination is generated. The path-time window combination includes a conflict-free global path formed by the node sequence and a planned time window corresponding to each node in the conflict-free global path. Distribute the global path-time window combination to each AGV; Based on the real-time position and speed of each AGV, the deviation between the actual node occupancy time of each AGV and the planned time window is calculated; If the deviation of any of the AGVs does not exceed a first deviation threshold, the AGV is instructed to adjust its travel speed; if the deviation of any of the AGVs exceeds the first deviation threshold, the path-time window combination is updated; or, it includes: The system receives a global path-time window combination sent by the server and runs based on its own path-time window combination. The path-time window combination includes a conflict-free global path formed by a sequence of nodes and a planned time window corresponding to each node in the conflict-free global path. When encountering obstacles, an incremental detour path is generated based on local environmental information; based on the estimated length of the detour path, the current travel speed, and the global path-time window combination, it is determined whether the detour path will cause new conflicts with other AGVs. If there is no conflict, the detour is carried out based on the detour path, and the updated path-time window combination is synchronized to the server. If there is a conflict, the local adjustment conflict event and obstacle information are reported to the server.

[0086] Furthermore, this application also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the workshop multi-AGV path planning method provided by the above methods, the method comprising: The path is discretized into a sequence of nodes, a global time window matrix is ​​constructed, and a path-time window combination is generated. The path-time window combination includes a conflict-free global path formed by the node sequence and a planned time window corresponding to each node in the conflict-free global path. Distribute the global path-time window combination to each AGV; Based on the real-time position and speed of each AGV, the deviation between the actual node occupancy time of each AGV and the planned time window is calculated; If the deviation of any of the AGVs does not exceed a first deviation threshold, the AGV is instructed to adjust its travel speed; if the deviation of any of the AGVs exceeds the first deviation threshold, the path-time window combination is updated; or, it includes: The system receives a global path-time window combination sent by the server and runs based on its own path-time window combination. The path-time window combination includes a conflict-free global path formed by a sequence of nodes and a planned time window corresponding to each node in the conflict-free global path. When encountering obstacles, an incremental detour path is generated based on local environmental information; based on the estimated length of the detour path, the current travel speed, and the global path-time window combination, it is determined whether the detour path will cause new conflicts with other AGVs. If there is no conflict, the detour is carried out based on the detour path, and the updated path-time window combination is synchronized to the server. If there is a conflict, the local adjustment conflict event and obstacle information are reported to the server.

[0087] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0088] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0089] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for multi-AGV path planning in a workshop, characterized in that, Applied to the server side, including: The path is discretized into a sequence of nodes, a global time window matrix is ​​constructed, and a path-time window combination is generated. The path-time window combination includes a conflict-free global path formed by the node sequence and a planned time window corresponding to each node in the conflict-free global path. Distribute the global path-time window combination to each AGV; Based on the real-time position and speed of each AGV, the deviation between the actual node occupancy time of each AGV and the planned time window is calculated; If the deviation of any AGV does not exceed the first deviation threshold, the AGV is instructed to adjust its travel speed; if the deviation of any AGV is greater than the first deviation threshold, the path-time window combination is updated.

2. The workshop multi-AGV path planning method according to claim 1, characterized in that, The generation path-time window combination includes: Conflict detection steps: Detect whether there are node conflicts or opposing conflicts among the AGVs; Conflict resolution steps: If there are node conflicts or opposing conflicts, the conflicting AGVs are sorted according to preset priorities, and the low-priority AGVs are shifted within a time window or replanned based on a preset path planning algorithm. Repeat the conflict detection step and the conflict resolution step until there are no conflicts in the path-time window combinations of all AGVs.

3. The workshop multi-AGV path planning method according to claim 2, characterized in that, The path planning algorithm employs an improved TROA algorithm, which is initialized based on the following steps: Determine the dimension of the solution space, and determine a prime number based on the dimension of the solution space; Based on the prime numbers, generate the optimal point vector; Based on the optimal point vector, a unit space point set is generated and mapped to a raster map. The starting point and ending point are forced to be consistent with the task objective, resulting in different paths from the starting point to the ending point.

4. The workshop multi-AGV path planning method according to claim 3, characterized in that, The improved TROA algorithm is optimized by iterative parameter optimization based on the following steps: In the TROA algorithm, the hunting step size, territory range, ambush probability, and random roaming probability are each established with a nonlinear decay function relative to the current iteration progress, where the current iteration progress is the ratio of the current iteration number to the maximum iteration number. Based on the aforementioned nonlinear decay functions, the hunting stride length, the territory range, the ambush probability, and the random roaming probability are updated respectively.

5. The workshop multi-AGV path planning method according to claim 4, characterized in that, The nonlinear decay function of the hunting step length is as follows: ； in, It is the initial hunting stride length. It is the current iteration number. It is the maximum number of iterations. It is the decay index. It is the first Hunting stride length in that era.

6. The workshop multi-AGV path planning method according to claim 4, characterized in that, The nonlinear decay function of the raid probability is as follows: ； in, It is the initial raid probability. It is the current iteration number. It is the maximum number of iterations. It is the decay index. It is the minimum probability of a surprise attack. It is the first The probability of a surprise attack during a certain era.

7. The workshop multi-AGV path planning method according to claim 2, characterized in that, The update of the path-time window combination includes: If the deviation of the AGV does not exceed the second deviation threshold, the planned time window of all subsequent nodes of the AGV is adjusted, and it is checked whether the adjusted path-time window combination causes new conflicts with other AGVs. If no new conflicts are caused, the adjusted path-time window combination is sent to the above AGV. The second deviation threshold is greater than the first deviation threshold. If the deviation of the AGV is greater than the second deviation threshold or the adjusted path-time window combination causes a new conflict with other AGVs, the path-time window combination of the affected AGVs is adjusted based on the conflict resolution step.

8. A method for multi-AGV path planning in a workshop, characterized in that, Applications in AGVs include: The system receives a global path-time window combination sent by the server and runs based on its own path-time window combination. The path-time window combination includes a conflict-free global path formed by a sequence of nodes and a planned time window corresponding to each node in the conflict-free global path. When encountering obstacles, an incremental detour path is generated based on local environmental information; based on the estimated length of the detour path, the current travel speed, and the global path-time window combination, it is determined whether the detour path will cause new conflicts with other AGVs. If there is no conflict, the detour is carried out based on the detour path, and the updated path-time window combination is synchronized to the server. If there is a conflict, the local adjustment conflict event and obstacle information are reported to the server.

9. A multi-AGV path planning device for a workshop, characterized in that, include: The global planning module is used to: discretize the path into a sequence of nodes, construct a global time window matrix, and generate a path-time window combination, wherein the path-time window combination includes a conflict-free global path formed by the node sequence and a planned time window corresponding to each node in the conflict-free global path; The path distribution module is used to distribute the global path-time window combination to each AGV. The real-time monitoring module is used to: calculate the deviation between the actual node occupancy time of each AGV and the planned time window based on the real-time position and speed of each AGV; The dynamic adjustment module is used to: instruct the AGV to adjust its travel speed if the deviation of any AGV does not exceed a first deviation threshold; and update the path-time window combination if the deviation of any AGV exceeds the first deviation threshold.

10. A multi-AGV path planning device for a workshop, characterized in that, include: The control module is used to: receive a global path-time window combination sent by the server, and run based on its own path-time window combination, wherein the path-time window combination includes a conflict-free global path formed by a sequence of nodes, and a planned time window corresponding to each node in the conflict-free global path. The real-time detour module is used to: generate an incremental detour path based on local environmental information when encountering obstacles; and determine whether the detour path will cause new conflicts with other AGVs based on the estimated length of the detour path, the current driving speed, and the global path-time window combination. The local obstacle avoidance module is used to: if there is no conflict, detour based on the detour path and synchronize the updated path-time window combination to the server. If a conflict occurs, the local adjustment conflict event and obstacle information are reported to the server.

Images