Multi-AGV (Automatic Guided Vehicle) robot path planning method and system involving multi-task allocation

By adopting the first-come-first-service proximity allocation principle and an improved A* algorithm in multi-AGV robot path planning, combining state detection and conflict search, the problem of long-term multi-AGV path planning is solved, and transportation efficiency and fluency are improved. Fast pickup and delivery of tin ingots is achieved.

CN119088021APending Publication Date: 2024-12-06YUNNAN TIN CO LTD TIN BRANCH +1
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411210489.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

In multi-task allocation, the path solution time of multi-AGV robot path planning is long, resulting in low transportation efficiency.

Method used

Adopting the first-come-first-service near-access allocation principle, assign tasks to AGVs, and planning paths through improved A* algorithms, setting the expansion direction of the raster to reduce conflicts, and using state detection and conflict search to split nodes to resolve conflicts.

Benefits of technology

The speed of multi-AGV path planning is accelerated, the conflicts between AGVs are reduced, transportation efficiency and fluency are improved, and the rapid pick-up and delivery of tin ingots are achieved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119088021A_ABST
    Figure CN119088021A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of path planning, in particular to a multi-AGV robot path planning method and system relating to multi-task allocation, and the method comprises the steps: building a map model of a transport warehouse, and setting the extension direction of a grid; assigning tasks to each AGV by adopting a first-arrival first-service nearest assignment principle to obtain a task plan; detecting the state of each AGV, calculating the effective cargo carrying capacity C of an idle AGV, and when the effective cargo carrying capacity C of a certain idle AGV is lower than a threshold value, modifying the state of the AGV to be busy, taking a charging area as a target point, and updating task planning; and according to the starting point and the target point of each AGV in task planning, an initial path is planned for each AGV by adopting an improved A * algorithm, conflict search is performed on the initial path, nodes are split to solve conflicts, and when no conflict exists in a path set, an overall path is output. According to the invention, pickup and delivery of goods can be accelerated, the number of conflicts among the AGVs is reduced, and the speed of multi-AGV path planning is accelerated.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of path planning, and more particularly to a multi-AGV robot path planning method involving multi-task allocation. Background Art

[0002] In the tin processing process, the transportation of tin ingots is often a simple and boring task. During transportation, if workers work for a long time, it may lead to a decrease in transportation efficiency and bring some production safety hazards. Unlike manual labor, AGV robots can work around the clock, which improves productivity. In addition, they also reduce the risk of errors and accidents and improve the safety of the storage environment. Using robots to replace manual labor to complete some simple and repetitive tasks can improve the factory's transportation efficiency and strengthen the safety of the production environment. This not only reduces the labor cost of the enterprise, but also has important significance for the development of productivity.

[0003] The transportation of tin ingots involves task allocation and path planning. The robot needs to plan the path according to the assigned task to transport the tin ingots to the designated location. Whenever there is a new task assignment, the robot will plan a new path. Moreover, compared with a single AGV robot, a multi-AGV robot system has higher work efficiency and better stability. Therefore, in order to improve the transportation efficiency of tin ingots, task allocation and multi-AGV robot path planning have become more important research directions.

[0004] However, in the current multi-task allocation, when planning the paths of multiple AGV robots, there is a technical problem that the path solution time is long, resulting in low transportation efficiency. Summary of the invention

[0005] In view of this, the present invention provides a multi-AGV robot path planning method and system involving multi-task allocation, which can speed up the picking and delivery of tin ingots, reduce the number of conflicts between AGVs, and speed up the multi-AGV path planning.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a multi-AGV robot path planning method involving multi-task allocation, comprising the following steps:

[0008] Establish a map model of the transport warehouse and set the expansion direction of the grid in the map model;

[0009] Adopt the principle of first-come, first-served and nearest allocation to assign tasks to each AGV and obtain task planning;

[0010] Detect whether each AGV is idle or busy, and calculate the effective cargo capacity C of the idle AGV. When the effective cargo capacity C of an idle AGV is lower than the threshold, change the AGV state to busy and use the charging area as the target point to update the task planning.

[0011] According to the starting point and target point of each AGV in the task planning, the improved A* algorithm is used to plan the initial path for each AGV, and the initial path is searched for conflicts and nodes are split to resolve conflicts. When there is no conflict in the path set, the overall path is output.

[0012] Furthermore, the map model includes a charging area and a standby area for the AGV, and a palletizing area and a temporary stacking area for the goods, wherein the charging area and the standby area are arranged on the left side of the palletizing area, and the standby area is on the right side of the charging area and close to the palletizing area; the overall path direction in the map model presents a cycle structure from the palletizing area to the temporary stacking area and then to the palletizing area, wherein the direction from the palletizing area to the temporary stacking area is the delivery path, and the direction from the temporary stacking area to the palletizing area is the pickup path; by setting the expansion direction of the grid, the search can only be performed along the specified direction when planning the path for the AGV.

[0013] Furthermore, the principle of first-come, first-served distribution is adopted to allocate picking tasks or stacking tasks to each AGV, wherein the allocation of picking tasks includes:

[0014] Arrange the idle AGVs from front to back according to the distance between them and the palletizing area, determine the priority queue of the idle AGVs, and give priority to the idle AGVs that arrive at the palletizing area first in the priority queue of the idle AGVs according to the principle of proximity to the nearest goods;

[0015] The allocation of stacking tasks includes:

[0016] Arrange the busy AGVs from front to back according to their distance from the temporary stacking area to determine the priority queue of the busy AGVs; and give priority to allocating the nearest idle stacking point to the frontmost robot in the priority queue of the busy AGVs according to the principle of proximity;

[0017] Whenever an AGV completes a picking task or a stacking task, the corresponding priority queue is updated.

[0018] Furthermore, the picking task includes three types of situations. The first type of situation is: if an AGV completes the stacking task and moves from the temporary stacking area to the palletizing area, and no task arrives in the palletizing area, if the AGV arrives at the entrance of the palletizing area and still has no task after waiting for a period of time, the AGV enters the standby area; if a new task arrives in front of the AGV during the process of entering the standby area, it will be assigned according to the principle of first come first served;

[0019] The second situation is: if an AGV is in the process of moving from the temporary stacking area to the palletizing area, and a task arrives in the palletizing area, the task will be assigned to the nearest area on a first-come, first-served basis.

[0020] The third situation is: if an AGV has a task arriving at a farther palletizing position while no task is arriving at a closer palletizing position during its journey from the temporary stacking area to the palletizing area, the farther task will be assigned to the AGV first. However, after passing the closer palletizing position, a task will arrive at the closer palletizing position. At this time, the AGV ignores the task at the palletizing position that it has passed and assigns the newly arrived task to the next priority AGV.

[0021] Furthermore, the stacking task includes three types of situations. The first type of situation is: if all the stacking points in the temporary stacking area are idle when an AGV arrives at the temporary stacking area, the AGV randomly selects an idle stacking point from the temporary stacking area or defaults to the first idle stacking point as the target stacking point;

[0022] The second situation is: if an AGV arrives at the temporary stacking area, there are goods in the temporary stacking area, but there are also free stacking points, then the AGV selects the nearest free stacking point as the target stacking point to stack the goods;

[0023] The third situation is: if an AGV arrives at the temporary stacking area, and there are goods at all the stacking points in the temporary stacking area, then the AGV waits at the first stacking point in the temporary stacking area, and the subsequent AGVs wait in sequence.

[0024] Furthermore, when the effective cargo carrying capacity C value of an idle AGV is lower than the threshold, after the AGV state is changed to busy state, the priority queues of the idle AGV and the busy AGV are updated accordingly, and only the idle or standby AGVs participate in the cargo picking task allocation;

[0025] The calculation formula of the effective cargo capacity C value is:

[0026] C=l 取 +k*l 送 +c

[0027] Among them, c represents the current power of AGV, k represents the weight of the cargo, and l 取 Indicates the distance from the current location of the AGV to the pickup location, l 送 Indicates the distance from the AGV's pickup location to the stacking location.

[0028] Furthermore, before planning the path for each AGV, the path search rules of the A* algorithm are improved, and the search direction of each grid is set, specifically including: setting attributes for each grid point in the charging area, the standby area and the path, the attributes including position (x, y) and travelable direction (D1, D2, D3, D4) = (T / F, T / F, T / F, T / F); wherein D1, D2, D3, D4 respectively represent the four directions in the four-neighborhood search of the AGV, T / F indicates that the grid points in the corresponding direction are accessible / inaccessible, and each grid includes at least one searchable direction.

[0029] Furthermore, the process of searching for conflicts on the initial path and splitting nodes to resolve the conflicts includes:

[0030] A binary tree is used to search for conflicts in the initial path. Each node on the binary tree is called a CT node. The conflict type of a non-target CT node is divided into point conflicts. i , a j , v, t> and edge conflicts i , a j ,e,t>, when generating child nodes, point conflicts i , a j , v, t> generates two constraints i , v, t> and j , v, t> is added to the two child nodes, indicating that the AGV robot a i or a j Cannot be at point v at time t, edge conflict i , a j , e, t> also generates two constraints i ,e,t> and j , e, t>, represents AGV robot a i or a j It cannot be on the e side at time t;

[0031] If there are still conflicts in the paths of the CT nodes, the child nodes will continue to be split to resolve the conflicts, and each split node will be added to the candidate node list until all conflicts are resolved.

[0032] Furthermore, the method for detecting the conflict between any two AGVs is as follows:

[0033] Assume that AGV robot a i The path is AGV robot j The path is

[0034] Get the path p through the function get() i The position of any point in​​​​​​​​ and time step And the path p j The position of any point in and time step

[0035]

[0036] Detect two paths p i and p j The method to determine whether there is a conflict is as follows:

[0037]

[0038] In a second aspect, the present invention provides a multi-AGV robot path planning system involving multi-task allocation, which is applicable to the multi-AGV robot path planning method involving multi-task allocation as described above, including:

[0039] A map modeling module is used to establish a map model of a transport warehouse and set the expansion direction of the grid in the map model;

[0040] The task allocation module is used to allocate tasks to each AGV based on the principle of first-come, first-served and nearest allocation, and obtain task planning;

[0041] The state detection module is used to detect whether the state of each AGV is idle or busy, and calculate the effective cargo capacity C of the idle AGV. When the effective cargo capacity C of an idle AGV is lower than the threshold, the state of the AGV is changed to busy and the charging area is used as the target point;

[0042] The initial path generation module is used to plan the initial path for each AGV using the A* algorithm according to the starting point and target point of each AGV in the task planning;

[0043] The conflict handling module is used to search for conflicts on the initial path and split nodes to resolve conflicts. When there is no conflict in the path set, the overall path is output.

[0044] It can be seen from the above technical solutions that, compared with the prior art, the present invention has the following beneficial effects:

[0045] The first-come, first-served principle of nearest distribution proposed in the present invention allocates the nearest goods to the front AGV robot. After receiving the goods, the AGV robot can move to the delivery route to the temporary stacking area, avoiding possible conflicts with subsequent picking robots. Similarly, when stacking goods, the goods are also stacked according to the first-come, first-served principle of nearest stacking. After the AGV robot completes the delivery, it can move to the pickup route.

[0046] The present invention sets the passable directions of grid points and plans paths for multiple robots based on a conflict search algorithm. When planning the paths, the robots can only search along the set directions, thereby reducing the number of conflicts and turns that occur when AGV robots are transporting goods. Goods with a large throughput can be transported in an orderly manner, the smoothness of transportation is improved, the picking and delivery of tin ingots is accelerated, and the overall transportation efficiency is improved.

[0047] During task planning, the present invention realizes robot status detection and control, including standby and charging, to ensure the smooth and orderly progress of each AGV robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0049] Figure 1 A flow chart of a multi-AGV robot path planning method involving multi-task allocation provided by the present invention;

[0050] Figure 2 A map model after setting the grid search direction provided by the present invention;

[0051] Figure 3(a)-Figure 3(c) They are schematic diagrams of adopting the principle of nearest allocation for picking up goods under different situations provided by the present invention;

[0052] Figure 4(a)-Figure 4(c) They are schematic diagrams of adopting the principle of nearest distribution for delivering goods under different situations provided by the present invention;

[0053] Figure 5(a)-Figure 5(f) They are respectively schematic diagrams of search directions of the path planning algorithm provided by the present invention;

[0054] Figure 6(a)-Figure 6(b) They are schematic diagrams of point conflict and edge conflict provided by the present invention respectively. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] like Figure 1As shown, an embodiment of the present invention discloses a multi-AGV robot path planning method involving multi-task allocation, comprising the following steps:

[0057] S1. Establish a map model of the transport warehouse and set the expansion direction of the grid in the map model;

[0058] S2, adopt the principle of first-come, first-served, and assign tasks to each AGV to obtain task planning;

[0059] S3, detect whether the status of each AGV is idle or busy, and calculate the effective cargo capacity C of the idle AGV. When the effective cargo capacity C of a certain idle AGV is lower than the threshold, change the status of the AGV to busy and use the charging area as the target point to update the task planning;

[0060] S4. According to the starting point and target point of each AGV in the task planning, the improved A* algorithm is used to plan the initial path for each AGV, and the initial path is searched for conflicts and nodes are split to resolve conflicts;

[0061] S5. When there is no conflict in the path set, output the entire path. If there is a conflict, return to S4 to continue resolving the conflict.

[0062] The above steps are further explained below.

[0063] S1. Establish a map model.

[0064] like Figure 2 As shown in the figure, the map model includes the charging area and standby area of ​​AGV, as well as the palletizing area and temporary stacking area of ​​goods. There are N production lines for tin ingot processing, and the number of palletizing areas and temporary stacking areas for the corresponding goods is also N. The palletizing area corresponds to the production line position one by one, and the temporary stacking area is set close to the transport vehicle that transports the goods to the corresponding workshop. Among them, the number of grids of the charging area and standby area of ​​AGV is M, which is set on the left side of the palletizing area of ​​the production line, and the standby area is on the right side of the charging area and close to the palletizing area. The overall path direction in the map model presents a cyclic structure from the palletizing area to the temporary stacking area and then to the palletizing area, where the direction from the palletizing area to the temporary stacking area is the delivery path, and the direction from the temporary stacking area to the palletizing area is the pickup path; by setting the expansion direction of the grid, the AGV can only search along the specified direction when planning the path. When the battery of the robot is reduced, it arrives at the charging area for charging.

[0065] During the experiment, the simulation program of the computer is used for implementation, the execution subject is the computer, and the development environment adopts the Linux system. The number of AGV robots and the number of production lines can be set by the user, and a certain throughput of goods is set to test the performance of the method of the present invention.

[0066] S2. Goods distribution.

[0067] The principle of picking up and stacking goods is first-come, first-served. The robots that arrive at the palletizing area and temporary stacking area first are given priority in allocating the nearest goods and the nearest stacking point. After picking up and stacking are completed, they reach the next target point according to the path planning results and the direction of travel specified by the grid to reduce the number of conflicts and speed up transportation efficiency.

[0068] Among them, the allocation of pickup tasks includes:

[0069] Arrange the idle AGVs from front to back according to the distance between them and the palletizing area, determine the priority queue of the idle AGVs, and give priority to the idle AGVs that arrive at the palletizing area first in the priority queue of the idle AGVs according to the principle of proximity to the nearest goods;

[0070] The allocation of stacking tasks includes:

[0071] Arrange the busy AGVs from front to back according to their distance from the temporary stacking area to determine the priority queue of the busy AGVs; and give priority to allocating the nearest idle stacking point to the frontmost robot in the priority queue of the busy AGVs according to the principle of proximity;

[0072] Whenever an AGV completes a picking task or a stacking task, the corresponding priority queue is updated.

[0073] Specifically, the steps for assigning tasks to each AGV include:

[0074] a. Set the priority queue of the idle AGV robot LeisureList = {a1, a2, ..., a i} and the priority queue of the pickup task PickupList = {t1, t2, ..., t i}, and accordingly set the priority queue of the busy AGV robot BusyList = {b1, b2, ..., b i} and the priority queue DeliveryList = {k1, k2, ..., k i}, each time the task is assigned starting from the first AGV robot in the list.

[0075] b. Determine the number of idle AGV robots and determine the distance between each idle robot and the palletizing area. Arrange the picking priority of each AGV robot from front to back according to position, and insert them into the priority queue LeisureList in sequence. When assigning tasks, assign tasks to the robots starting from the head of the queue.

[0076] c. Read the quantity of goods to be picked up, and arrange the priority of each picking task from near to far according to the location, and insert them into the priority queue PickupList in sequence. When picking up goods, start picking up and distributing them from the head of the queue; that is, assign goods t1 to robot a1, assign goods t2 to robot a2, and so on.

[0077] d. Accordingly, the busy AGV robots (excluding charging robots) are arranged from front to back according to the distance from the temporary stacking area, and are inserted into the priority queue BusyList of the busy AGV robots in sequence. The idle stacking points are arranged from near to far according to their positions, and are inserted into the priority queue DeliveryList of the stacking tasks in sequence. When stacking goods, the goods are stacked from the head of the busy robot queue BusyList to the nearest grid stacking point in the DeliveryList queue.

[0078] e. Whenever the AGV robot completes picking up or stacking, it updates the priority queue to ensure the accuracy of the principle of allocating goods nearby.

[0079] Among them, the picking task includes three types of situations. The first type of situation is: if an AGV completes the stacking task and moves from the temporary stacking area to the palletizing area, there is no task arriving in the palletizing area. If the AGV arrives at the entrance of the palletizing area and still has no task after waiting for a period of time, the AGV enters the standby area to wait; as shown in Figure 3(a), if there is no task arriving in the palletizing area, the robot will move to the standby area after waiting for a period of time at the entrance of the palletizing area. If a new task arrives in front of the AGV during its entry into the standby area, it will be assigned according to the principle of first come first served.

[0080] The second situation is: if a task arrives in the palletizing area during the process of an AGV moving from the temporary storage area to the palletizing area, the first-come-first-served principle is followed in this case; as shown in Figure 3(b), according to the first-come-first-served principle, the priority order of the robots in LeisureList is {a1, a2, ...,}. According to the priority, the robot a1 is assigned tasks first, and according to the principle of proximity, task a is assigned to robot No. 1, and task b is assigned to robot No. 2.

[0081] The third situation is: if a task arrives at a distant palletizing position while an AGV is moving from the temporary stacking area to the palletizing area, but no task arrives at a closer palletizing position, the distant task will be assigned to the AGV first. However, after passing the closer palletizing position, a task arrives at the closer palletizing position. At this time, the AGV ignores the task at the palletizing position it has passed and can only continue to execute the previous task, assigning the newly arrived task to the AGV with the next highest priority. As shown in Figure 3(c), when robot No. 1 passes the first position in the palletizing area, task b has not yet been reached, so task a is assigned to robot No. 1. When robot No. 1 reaches the third position in the palletizing area, task b is reached, and task b is assigned to robot No. 2.

[0082] Specifically, the farther palletizing area and the closer palletizing area refer to areas that the robot has walked through, the farther area is an area that the robot has not reached, and the closer area is an area that the robot has walked through.

[0083] The stacking task also includes three types of situations. The first type of situation is: if all the stacking points in the temporary stacking area are idle when an AGV arrives at the temporary stacking area, the AGV randomly selects an idle stacking point from the temporary stacking area or defaults to the first idle stacking point as the target stacking point; each stacking point is represented by a grid. As shown in Figure 4(a), when the temporary stacking area is empty, the first grid is selected as the target point by default, and the robot stacks the goods at the first grid point.

[0084] The second situation is: if an AGV arrives at the temporary stacking area, there are goods in the temporary stacking area, but there are also free stacking points, then the AGV selects the nearest free stacking point as the target stacking point to stack the goods; as shown in Figure 4(b), robot No. 1 selects position a as the target point nearby. After robot No. 1 completes the task assignment, the stacking position closest to robot No. 2 is position b, so position b is assigned to it as the target point.

[0085] The third situation is: if an AGV arrives at the temporary stacking area, and all the stacking points in the temporary stacking area have goods, then the AGV waits at the first stacking point in the temporary stacking area, and the subsequent AGVs wait in sequence. As shown in Figure 4(c), robots 1 and 2 are waiting at the entrance of the temporary stacking area until the goods are taken away by the next workshop and the temporary stacking area is vacated before stacking tasks are assigned.

[0086] Among them, the state control of the robot is set in the parameters of the simulation program class object. By reading and writing the corresponding parameters, the state of the robot can be controlled.

[0087] In general, when the AGV robot is performing picking and stacking tasks, it needs to perform different operations when encountering different situations, including entering the standby area; ignoring the arrival of new goods in the area that has been passed; randomly selecting a grid for stacking when there is no goods in all grids in the temporary stacking area; selecting the nearest grid for stacking when there are goods in some grids in the temporary stacking area; and waiting when there are goods in all grids in the temporary stacking area.

[0088] S3. Detect the status of each AGV.

[0089] The present invention sets a central detection unit to detect whether each AGV robot is idle or busy. When it is idle, tasks can be assigned to the corresponding robot. In order to avoid the problem that the robot slows down or even stops due to insufficient power during transportation, the present invention sets the effective cargo capacity of the robot. When the effective cargo capacity C value of a certain idle AGV is lower than the threshold, the robot state is set to busy and the charging area is used as the target point. At the same time, the priority queues of the idle AGV and the busy AGV are updated accordingly. Only the AGVs in the idle or standby state participate in the task allocation of picking up goods.

[0090] The calculation formula of the effective cargo capacity C value is:

[0091] C=l 取 +k*l 送 +c

[0092] Among them, c represents the current power of AGV, k represents the weight of the cargo, and l 取 Indicates the distance from the current location of the AGV to the pickup location, l 送 Indicates the distance from the AGV's pickup location to the stacking location.

[0093] If the robot is in the standby area and no tasks arrive for a period of time, the state changes to standby and enters energy saving mode.

[0094] S4. Initial path planning and conflict resolution.

[0095] S401, map preprocessing:

[0096] Before path planning, the map needs to be preprocessed. In traditional grid maps, the path search direction of the robot is mostly a 4-neighborhood search. In the map of the present invention, the search direction of the grid points is set to form the pickup path and the delivery path, and the robot is required to search only along the set direction.

[0097] Specifically, set attributes for each grid point in the charging area, standby area, and path. The attributes include position (x, y) and travelable direction (D1, D2, D3, D4) = (T / F, T / F, T / F, T / F); where D1, D2, D3, and D4 represent the four directions in the AGV four-neighborhood search, and T / F represents the grid point in the corresponding direction is accessible / inaccessible. When the value is T, it can be expanded, and when it is F, it cannot be expanded. Each grid includes at least one searchable direction. As shown in Figure 5(a), the robot's expansion direction defaults to 4 neighborhoods, and Figure 5(b) to Figure 5(f) The grid in the middle has a specified search direction, and the red direction indicates that the robot cannot access it. As shown in Figure 5(b), the robot can only expand along the D2 direction.

[0098] The steps to set the grid search direction are: first, default the search direction of all grid points to 4 neighborhoods; read the map, determine the position of each grid, traverse all grids, compare with the pre-given map model, determine the coordinates of the grid points and trim the directions that cannot be expanded until all grids are traversed.

[0099] S402: Loading a map.

[0100] The map can be represented by G = (V, E), where V is each grid in the map and E is the edge between grids. The present invention adds (D1, D2, D3, D4) to the attributes of the grid. When planning a path, the algorithm can only search and expand along the specified direction. After setting up the map, the map data is loaded into the reading entry of the path planning algorithm for solving.

[0101] S403, loading the priority queue for picking and stacking. The priority of picking and stacking is also stored in a circular queue. The tasks to be completed are taken from the head of the queue and delivered to the robot, and the tasks are converted into starting points and target points. After the robot completes a task, it is immediately imported into the next target point for subsequent operations.

[0102] S404, assign the goods to be transported and stacked to the robots. First, determine the number of idle robots, and load the robot index into the pickup order priority queue according to the position of each idle robot. Load the arrived tasks into the task waiting area. Task allocation adopts the principle of first come first served. Each task always selects the first robot from the pickup priority queue for transportation, that is, selects the nearest robot for transportation. The stacking task selects the nearest position from the temporary stacking area as the target point of the first robot in the stacking priority queue.

[0103] Also load the robot's index into the stacking order priority queue based on the position information. If the robot is in the charging state, ignore it.

[0104] S405, import the starting point and target point of the robot. After the task is assigned, it is converted into the starting point and target point and imported into the solution entry of the planning algorithm. Determine whether the robot's target point is accessible. If it is accessible, plan the path. If it is not accessible, reallocate the task. Monitor the target point during the path planning process. For the temporary stacking area, if the stacked goods have not been taken away by the next workshop, wait or select other grid points as the stacking target point.

[0105] S406, solving the initial path of each robot. When planning the initial path, other robots are ignored and conflict constraints are not considered. The process of planning the initial path is carried out in the root node of the binary tree. When expanding the node, the search can only be carried out along the direction specified by the grid.

[0106] The multi-robot path planning method in the present invention is improved on the basis of the conflict-based search (CBS) algorithm. The conflict-based search algorithm consists of two parts, namely, a conflict handling part and a path planning part, wherein the path planning adopts the A* algorithm. The present invention improves the search rules of the path planning part, sets the search direction of each grid, and the improved A* algorithm is called the IA* (Improved A*) algorithm. When the IA* algorithm searches, the nodes in the Open list can only be expanded along a specific direction. The planning path first ignores other robots, and uses the improved A* algorithm to solve the initial path for all robots in turn. The initial path may contain conflicts, which are resolved by splitting nodes in the subsequent conflict handling part.

[0107] S407. Conflict resolution.

[0108] After planning the initial path, conflicts between robots on the path need to be resolved to output the final valid path. Conflict constraints need to be considered when planning a new path so that conflicts between robots can be resolved, and conflict constraints in all parent nodes need to be considered.

[0109] The algorithm framework based on conflict search consists of a binary tree. Each node on the binary tree is called a CT node. The paths of all robots are compared two by two. If there are identical parts in time and space between two paths, it means that there is a conflict between the two robots. The corresponding conflict constraint is added to the CT child node, and the conflict is avoided when planning a new path.

[0110] Considering a non-target CT node, the conflict type is divided into point conflict i , a j ,v,t> and edge conflicts i , a j ​​,e,t>, when generating child nodes, point conflicts i , a j , v, t> generates two constraints i , v, t> and j , v, t> is added to the two child nodes, indicating that the AGV robot a i or a j Cannot be at point v at time t, edge conflict i , a j , e, t> also generates two constraints i ,e,t> and j , e, t>, represents AGV robot a i or a j It cannot be on the e edge at time t; this constrains the path planning of the A* algorithm. The robot with the conflict constraint calls the IA* algorithm to plan a new path. When planning the path, the robot is restricted to access the specified grid at a specific time (similar to setting a temporary obstacle) to avoid conflicts with other robots.

[0111] If there are still conflicts in the paths of the CT nodes, the child nodes will continue to be split to resolve the conflicts, and each split node will be added to the candidate node list until all conflicts are resolved.

[0112] The search steps of the IA* algorithm with added conflict constraints are as follows:

[0113] a. Read the starting point and target point, and put the starting point into the Open list;

[0114] b. Select the node N with the smallest replacement value f from the Open list for expansion;

[0115] c. Determine whether the selected node is the target point, if so, the search is completed;

[0116] d. Determine the scalable direction of the node and generate child nodes N c And calculate the cost value f(N c )=g(N c )+h(N c ), then the node N c Put them into the Open list and sort them from small to large according to the cost value;

[0117] e. Put the expanded node N into the Close list;

[0118] f. Return to step b until the target point is found or the Open list is empty.

[0119] The way to detect the conflict between any two AGVs is: ​​​​​​

[0120] Assume that AGV robot a i The path is AGV robot j The path is

[0121] Get the path p through the function get() i The position of any point in and time step And the path p j The position of any point in and time step

[0122]

[0123] Detect two paths p i and p j The method to determine whether there is a conflict is as follows:

[0124]

[0125] S408, output path. Detect the paths of all robots through conflict search. When there is no conflict, output the paths of all robots, and the algorithm search ends.

[0126] During path planning, congestion detection is also required. When the goods in the temporary storage area have not been taken away for a long time, the robot transporting the tin ingots will be congested near the temporary storage area. At this time, the algorithm stops planning the path, retains the existing path results, and makes the robot wait. When the goods in the temporary storage area are taken away, the algorithm is restarted for path planning.

[0127] In general, the steps of path planning and conflict resolution for multiple robots in the present invention are as follows: (1) Import the starting point and target point of all robots; (2) Plan the initial paths of all robots from the starting point to the target point in sequence in the root node of the binary tree; (3) Detect conflicts and compare the paths of all robots in pairs. As shown in Figure 6(a), point conflict refers to the existence of parts that are identical in time and space between two paths. As shown in Figure 6(b), edge conflict refers to the swapping of the positions of two robots at the previous and next moments. (4) Split the CT node, add conflict constraints, and call the IA* algorithm to plan a new path to resolve the conflict. (5) Continuously split the CT node until there is no conflict in the node, then output a valid path.

[0128] Next, the performance of the present invention is verified by combining experimental data:

[0129] Map adoption Figure 2In the map, the number of robots increases from 6 to 10, the number of production lines increases from 16 to 20, and the number of tasks increases from 120 to 200. It is assumed that the goods will be picked up by the next workshop after five moments in the temporary stacking area. By comparing the PBS (Priority-Based Search, PBS) algorithm, the performance of the proposed method is analyzed from the program running time Runtime (seconds). The method proposed in this invention is the CBS-PD (CBS for Pickup and Delivery) algorithm. The test results are shown in Table 1.

[0130] Table 1

[0131]

[0132] It can be seen from the above experiments that the method of the present invention can obtain a faster solution time in the test map, thereby accelerating the transportation efficiency of tin ingots.

[0133] In other embodiments, the present invention further provides a multi-AGV robot path planning system involving multi-task allocation, which is applicable to the multi-AGV robot path planning method involving multi-task allocation as above, including:

[0134] A map modeling module is used to establish a map model of a transport warehouse and set the expansion direction of the grid in the map model;

[0135] The task allocation module is used to allocate tasks to each AGV based on the principle of first-come, first-served and nearest allocation, and obtain task planning;

[0136] The state detection module is used to detect whether the state of each AGV is idle or busy, and calculate the effective cargo capacity C of the idle AGV. When the effective cargo capacity C of an idle AGV is lower than the threshold, the state of the AGV is changed to busy and the charging area is used as the target point;

[0137] The initial path generation module is used to plan the initial path for each AGV using the A* algorithm according to the starting point and target point of each AGV in the task planning;

[0138] The conflict handling module is used to search for conflicts on the initial path and split nodes to resolve conflicts. When there is no conflict in the path set, the overall path is output.

[0139] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0140] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-AGV robot path planning method involving multi-task allocation, characterized in that: The following steps are involved: Establish a map model of the transport warehouse and set the expansion direction of the grid in the map model; Adopt the principle of first-come, first-served and nearest allocation to assign tasks to each AGV and obtain task planning; Detect whether each AGV is idle or busy, and calculate the effective cargo capacity C of the idle AGV. When the effective cargo capacity C of an idle AGV is lower than the threshold, change the AGV state to busy and use the charging area as the target point to update the task planning. According to the starting point and target point of each AGV in the task planning, the improved A* algorithm is used to plan the initial path for each AGV, and the initial path is searched for conflicts and nodes are split to resolve conflicts. When there is no conflict in the path set, the overall path is output.

2. The multi-AGV robot path planning method involving multi-task allocation according to claim 1 is characterized in that: The map model includes the charging area and standby area of ​​the AGV, and the palletizing area and temporary stacking area of ​​the goods, wherein the charging area and the standby area are arranged on the left side of the palletizing area, and the standby area is on the right side of the charging area and close to the palletizing area; the overall path direction in the map model presents a cycle structure from the palletizing area to the temporary stacking area and then to the palletizing area, wherein the direction from the palletizing area to the temporary stacking area is the delivery path, and the direction from the temporary stacking area to the palletizing area is the pickup path; by setting the expansion direction of the grid, the AGV can only search along the specified direction when planning the path.

3. The multi-AGV robot path planning method involving multi-task allocation according to claim 2 is characterized in that: The principle of first-come, first-served distribution is adopted to assign picking tasks or stacking tasks to each AGV. The picking task allocation includes: Arrange the idle AGVs from front to back according to the distance between them and the palletizing area, determine the priority queue of the idle AGVs, and give priority to the idle AGVs that arrive at the palletizing area first in the priority queue of the idle AGVs according to the principle of proximity to the nearest goods; The allocation of stacking tasks includes: Arrange the busy AGVs from front to back according to their distance from the temporary stacking area to determine the priority queue of the busy AGVs; and give priority to allocating the nearest idle stacking point to the frontmost robot in the priority queue of the busy AGVs according to the principle of proximity; Whenever an AGV completes a picking task or a stacking task, the corresponding priority queue is updated.

4. The multi-AGV robot path planning method involving multi-task allocation according to claim 3 is characterized in that: There are three types of picking tasks. The first type is: if an AGV completes the stacking task and moves from the temporary stacking area to the palletizing area, and no task arrives in the palletizing area, if the AGV arrives at the entrance of the palletizing area and still has no task after waiting for a period of time, the AGV enters the standby area; if a new task arrives in front of the AGV during the process of entering the standby area, it will be assigned according to the principle of first come first served; The second situation is: if an AGV is in the process of moving from the temporary stacking area to the palletizing area, and a task arrives in the palletizing area, the task will be assigned to the nearest area on a first-come, first-served basis. The third situation is: if an AGV has a task arriving at a farther palletizing position while no task is arriving at a closer palletizing position during its journey from the temporary stacking area to the palletizing area, the farther task will be assigned to the AGV first. However, after passing the closer palletizing position, a task will arrive at the closer palletizing position. At this time, the AGV ignores the task at the palletizing position that it has passed and assigns the newly arrived task to the next priority AGV.

5. The multi-AGV robot path planning method involving multi-task allocation according to claim 3 is characterized in that: The stacking task includes three types of situations. The first type of situation is: when an AGV arrives at the temporary stacking area, all the stacking points in the temporary stacking area are idle, then the AGV randomly selects an idle stacking point from the temporary stacking area or defaults to the first idle stacking point as the target stacking point; The second situation is: if an AGV arrives at the temporary stacking area, there are goods in the temporary stacking area, but there are also free stacking points, then the AGV selects the nearest free stacking point as the target stacking point to stack the goods; The third situation is: if an AGV arrives at the temporary stacking area, and there are goods at all the stacking points in the temporary stacking area, then the AGV waits at the first stacking point in the temporary stacking area, and the subsequent AGVs wait in sequence.

6. The multi-AGV robot path planning method involving multi-task allocation according to claim 3 is characterized in that: When the effective cargo carrying capacity C value of an idle AGV is lower than the threshold, the AGV state is changed to busy state, and the priority queues of idle AGV and busy AGV are updated accordingly. Only AGVs in idle or standby state participate in the task allocation of picking up goods; The calculation formula of the effective cargo capacity C value is: C=l 取 +k*l 送 +c Among them, c represents the current power of AGV, k represents the weight of the cargo, and l 取 Indicates the distance from the current location of the AGV to the pickup location, l 送 Indicates the distance from the AGV's pickup location to the stacking location.

7. The multi-AGV robot path planning method involving multi-task allocation according to claim 2 is characterized in that: Before planning the path for each AGV, the path search rules of the A* algorithm are improved and the search direction of each grid is set, including: setting attributes for each grid point in the charging area, standby area and path, the attributes include position (x, y) and travelable direction (D1, D2, D3, D4) = (T / F, T / F, T / F, T / F); where D1, D2, D3, and D4 represent the four directions in the AGV four-neighborhood search, respectively, and T / F indicates that the grid points in the corresponding direction are accessible / inaccessible, and each grid includes at least one searchable direction.

8. The multi-AGV robot path planning method involving multi-task allocation according to claim 7 is characterized in that: The process of searching for conflicts on the initial path and splitting nodes to resolve conflicts includes: A binary tree is used to search for conflicts in the initial path. Each node on the binary tree is called a CT node. The conflict type of a non-target CT node is divided into point conflicts. i ,a j ,v,t> and edge conflicts i ,a j ,e,t>, when generating child nodes, point conflicts i ,a j ,v,t> generates two constraints i ,v,t> and j ,v,t> is added to the two child nodes, indicating that the AGV robot a i or a j Cannot be at point v at time t, edge conflict < a i , a j , e, t> also generates two constraints <a i ,e,t> and j , e, t>, represents AGV robot a i or a j It cannot be on the e side at time t;​​​​​​ If there are still conflicts in the paths of the CT nodes, the child nodes will continue to be split to resolve the conflicts, and each split node will be added to the candidate node list until all conflicts are resolved.

9. The multi-AGV robot path planning method involving multi-task allocation according to claim 1, characterized in that: The way to detect the conflict between any two AGVs is: Assume that AGV robot a i The path is AGV robot j The path is Get the path p through the function get() i The position of any point in and time step And the path p j The position of any point in and time step Detect two paths p i and p j The method to determine whether there is a conflict is as follows:

10. A multi-AGV robot path planning system involving multi-task allocation, characterized in that: It is applicable to the multi-AGV robot path planning method involving multi-task allocation as described in any one of claims 1 to 9, comprising: A map modeling module is used to establish a map model of a transport warehouse and set the expansion direction of the grid in the map model; The task allocation module is used to allocate tasks to each AGV based on the principle of first-come, first-served and nearest allocation, and obtain task planning; The state detection module is used to detect whether the state of each AGV is idle or busy, and calculate the effective cargo capacity C of the idle AGV. When the effective cargo capacity C of an idle AGV is lower than the threshold, the state of the AGV is changed to busy and the charging area is used as the target point; The initial path generation module is used to plan the initial path for each AGV using the A* algorithm according to the starting point and target point of each AGV in the task planning; The conflict handling module is used to search for conflicts on the initial path and split nodes to resolve conflicts. When there is no conflict in the path set, the overall path is output.

Citation Information

Cited By

  • Multi-AGV cooperative path planning and scheduling method for chip intelligent storage

    CN120355050A

  • Warehouse AGV path planning method based on improved wolf pack algorithm

    CN120578167A