A path optimization method for real-time scheduling of multiple AGVs based on time window

Abstract

The present invention discloses a path optimization method for real-time scheduling of multiple AGVs based on a time window. First, by collecting the terrain environment information of the unmanned workshop, an unmanned workshop model is constructed. The system receives the material request from the machine tool, allocates idle trolleys, and uses the A* algorithm or Dijkstra algorithm to achieve path planning and add time window marks. The routes of the working trolleys are traversed one by one, and the time window check method is used to judge whether there is a time conflict and the type of conflict. According to the type of conflict, the conflict is eliminated and the route is optimized for the second time. Through the real-time scheduling method that combines route planning and conflict resolution, when statically allocating routes to idle trolleys, it can predict in advance that there will be a conflict with the working trolleys at a certain time point in the future, and can avoid the conflict at a certain time point in advance, effectively improving the efficiency of transporting materials, providing a new way and optimization method for system scheduling control, and having broad application prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of navigation and control, and particularly relates to a path optimization method for real-time scheduling of multiple AGVs based on a time window. Background Art

[0002] Automated Guided Vehicle (AGV) is one of the key devices in modern industrial automation logistics systems. AGVs have remarkable characteristics such as flexibility and intelligence, and can conveniently reorganize the system to meet the requirements of flexible transportation in manufacturing. Compared with traditional manual or semi-manual material transportation methods, AGV systems reduce labor intensity, reduce potential hazards during material transportation, and greatly improve production efficiency, playing an important role in various industries.

[0003] For the path planning of traditional AGV cars, according to the environmental impact of a factory or workshop without workers, a conflict-free and highly efficient path is planned for each AGV car through laser guidance technology. The path planning of AGV cars mainly includes three aspects of problems: (1) determining whether there is a feasible path between the starting point and the target point. (2) The path planned for the AGV car must be unblocked, conflict-free, and deadlock-free. (3) The path planned for the AGV car should make the operation efficiency of the entire system reach a relatively optimal effect. Among them, laser guidance technology is mainly divided into two parts: the AGV laser scanner and the AGV reflector. The laser scanner installed on the AGV car rotates 360° at a fixed speed and emits laser light towards the reflector. The laser scanner can detect the position of the reflector and obtain the information of the laser angle according to the direction of the laser light returned by the reflector. The AGV on-vehicle computer receives the above information and processes it to calculate the position and movement direction of the AGV car. During the operation of the AGV car, it sends its current position to the upper control system, and the upper control system compares and corrects the position of the AGV car with the parameters of this AGV car built into the system, so as to guide the AGV car to drive along the correct route. For occasions that require multiple AGV cars, there is a problem of scheduling multiple AGVs.

[0004] At present, the scheduling methods that are widely applied and studied mainly include scheduling algorithms based on time-window path planning, scheduling algorithms based on conflict avoidance, scheduling algorithms based on artificial intelligence prediction, etc. The scheduling algorithm based on time-window path planning is a simple and practical scheduling method that can avoid conflicts and waiting deadlocks among AGV cars. This method stipulates that in a certain time period, a certain section of the road is occupied by a certain AGV car, and other AGV cars are not allowed to drive into this section of the road during this time period; based on the conflict types of time windows, they are generally divided into concurrent conflict types and head-on conflict types; the global path planning method based on time windows is mainly used for static global path planning, attempting to solve the problem of avoiding conflicts. This method is very time-consuming to solve, especially when there are already many path constraints, and even a suitable path cannot be obtained. Peng Chengji's paper "Research on Multi-AGV Path Conflicts Based on Time Window Algorithm" in the "Proceedings of the Excellent Papers of the Academic Annual Conference of the Chinese Tobacco Society" proposed a dynamic path planning method based on time windows. It pauses at the conflict node and then calls the Dijstra algorithm again to re-plan the subsequent path. This method pauses at the conflict node or moves to the obstacle avoidance point, then modifies the time window of each node in the subsequent path, and re-checks whether there are conflicts in the nodes with the modified time windows. If there are conflicts, this conflict is repeatedly resolved until all paths are checked. The disadvantage of this method is that the re-planned path may still be the original path, resulting in the failure of the planning.

[0005] For example, the invention with the patent number CN113515117A proposes a conflict resolution method for multi-AGV real-time scheduling based on time windows. The method for dealing with head-on conflicts is to add a let-site S after the conflict node, and then the time stamps of all departure points are increased by 1. The disadvantage of this method is that there is no calculation of the stop-and-wait time after entering the let-site. After the car enters the let-site S and returns to the original path, there may still be head-on conflicts, and there is no secondary optimization of the path after the conflict to avoid conflicts again. Summary of the Invention

[0006] In view of the deficiencies in the prior art, the present invention provides a path optimization method for multi-AGV real-time scheduling based on time windows. By digitizing the terrain environment information, an unmanned workshop model is established, and at the same time, a real-time scheduling method that combines route planning and conflict resolution. When statically allocating routes to idle cars, it can be predicted in advance that there will be conflicts with working cars at a certain time point in the future. Therefore, the route can be optimized again when allocating the route, which will avoid conflicts at a certain time point in the future in advance and effectively improve the efficiency of transporting materials.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] An embodiment of the present invention proposes a path optimization method for real-time scheduling of multiple AGVs based on a time window. The method includes the following steps:

[0009] S1. Collect the topographic environment information of the unmanned workshop, perform data grid processing, and construct an unmanned workshop model;

[0010] S2. Receive the machine tool material request, allocate idle trolleys, and combine with the unmanned workshop model. Use the A* algorithm or Dijkstra algorithm to perform initial path planning for the machine tool material request, and add corresponding time window marks; the machine tool material request includes a source point and a target node;

[0011] S3. Traverse the routes of the working AGV trolleys one by one, and combine with the time window to determine whether there is a time conflict between the unfinished routes and the initial path planned in step S2. If there is no conflict, add the route to the route set P, and go to step S6. If there is a conflict, further determine the conflict type and transfer to step S4; the conflict types include co-point conflict and head-on conflict;

[0012] S4. According to the conflict type, with the goal of minimizing the number of stop-and-wait times, eliminate the conflict by making one of the AGV trolleys stop and wait, and perform secondary optimization on the initial path;

[0013] S5. Refresh the route set P according to the preset scheduling period, and eliminate the routes that each AGV trolley has already traveled, the route departure points, and the time window marks of the corresponding departure points;

[0014] S6. Repeat steps S2 to S5 until the scheduling of all AGV trolleys is completed.

[0015] Further, in step S2, the process of performing initial path planning for the machine tool material request and adding corresponding time window marks includes the following sub-steps:

[0016] S21. Receive the machine tool material request, combine with the unmanned workshop model, and according to the position of the machine tool, use the A* algorithm or Dijkstra algorithm to obtain an optimal path L a ={r 1 ,r 2 ,r 3 ,...r N}, where N is the number of path nodes of this path, and assign this path to an idle AGV trolley a;

[0017] S22. According to the departure time t of the trolley a, add a time stamp to each departure point in the path L a to obtain the time stamp set T a ={t 1 ,t 2 ,t 3,...t N}, where t N is the start time of t and the time to reach the Nth node r N .

[0018] S23. Add the scheduling route set and route timestamp set of this path to the route set P. The specific representation form is as follows:

[0019] p = p ∪ {L a , T a}}.

[0020] Furthermore, in step S3, the process of combining the time window to determine whether there is a time conflict between the unfinished route and the initial path planned in step S2 includes the following sub-steps:

[0021] Traverse the travel points r a in the line L i one by one, where i = 1, 2,..., N, and determine whether there is a time conflict with any travel point of the planned route in the route set P:

[0022] If it is detected that the path node of trolley a is the same as the path node of trolley b, and the timestamp corresponding to the path node of trolley a is the same as the timestamp corresponding to the path node of trolley b is the same as the previous path node of trolley b, and there is an intersection in the timestamps of the corresponding path nodes, then it is determined that there is an oncoming conflict between trolley a and trolley b.

[0023] Furthermore, in step S4, according to the conflict type, with the goal of minimizing the number of stop-and-wait times, the process of eliminating the conflict by making one of the AGV trolleys stop and wait and performing a secondary optimization of the initial path includes the following sub-steps:

[0024] When the conflict type is a common point conflict, mark the path node r i of the common point conflict. Add the path node r i+1 to one of the trolley's path nodes r i . The nodes after the original path node r i are postponed to the next node position, so that the trolley is at the path node r iPause once, and then modify the corresponding set of timestamps, t i+1 = t i + 1, and for all path nodes corresponding to the modified r i+1 increment the corresponding timestamps by 1;

[0025] When the conflict type is a head-on conflict, mark the entry path node r of the head-on conflict i , select the path node r of one of the vehicles i+1 as the yielding point S, and let the vehicle wait at the yielding point S until the vehicle in the head-on conflict drives away from the entry path node r i ; calculate the time t1 required for the selected vehicle to reach the yielding point S from the entry path node r i , and modify the corresponding timestamps Based on the distance from the current position of the vehicle in the head-on conflict to the entry path node r i , and the speed of the vehicle in the head-on conflict, calculate the waiting time t2 of the selected vehicle at the yielding point S; when the waiting time ends, the selected vehicle returns to the entry path node r of the head-on conflict i , modify the route r i+2 = r i , and for the corresponding timestamps after that, all are incremented by t1 + t2, where j ∈ {2, 3, 4,... N - i + 2}.

[0026] The beneficial effects of the present invention are as follows:

[0027] First, the path optimization method for real-time scheduling of multiple AGVs based on time windows proposed by the present invention, by digitizing the terrain environment information and establishing an unmanned workshop model, can be applied to the vast majority of unmanned workshops, port terminals, warehousing logistics and other environments, solves the limitations of automated guided vehicles (AGVs) in logistics operations, and expands more general scenarios.

[0028] Second, the path optimization method for real-time scheduling of multiple AGVs based on time windows proposed by the present invention proposes a real-time scheduling method that combines route planning and conflict resolution. When statically allocating routes to idle vehicles, it can predict in advance that there will be a conflict with a working vehicle at a certain time point in the future. Therefore, when allocating routes, the route can be optimized again to avoid conflicts at a certain time point in the future in advance.

[0029] Third, the path optimization method for real-time scheduling of multiple AGVs based on time windows proposed by the present invention, when a conflict occurs, calls the resolution strategy according to the conflict type to solve the current conflict, performs secondary optimization on the path after the conflict, finds a suitable obstacle avoidance point for waiting, greatly improves the operation efficiency of AGVs, and reduces the operation of redundant and ineffective routes.

[0030] Fourth, the path optimization method for real-time scheduling of multiple AGVs based on time windows proposed by the present invention can effectively avoid irresolvable multiple AGV conflict situations by setting specific obstacle avoidance points in complex situations, effectively improve the efficiency of material handling, provide new ways and optimization methods for system scheduling control, and have broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is a schematic flow chart of the path optimization method for real-time scheduling of multiple AGVs based on time windows according to an embodiment of the present invention.

[0032] Figure 2 is a schematic diagram of the map modeling structure of the unmanned workshop according to an embodiment of the present invention.

[0033] Figure 3 is a schematic diagram of the calculation method for the shortest path using the A* algorithm according to an embodiment of the present invention.

[0034] Figure 4 is a schematic diagram of the shortest path annotation according to an embodiment of the present invention.

[0035] Figure 5 is a schematic diagram of multiple AGV conflict cases according to an embodiment of the present invention.

[0036] Figure 6 is a schematic diagram of conflict resolution operation according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] Now, the present invention will be further described in detail with reference to the accompanying drawings.

[0038] It should be noted that the terms such as "upper", "lower", "left", "right", "front", "rear", etc. cited in the invention are only for the convenience of description and are not used to limit the scope of implementation of the present invention. The change or adjustment of their relative relationship shall be regarded as the scope of implementation of the present invention without substantial change in technical content.

[0039] Figure 1 is a schematic flow chart of the path optimization method for real-time scheduling of multiple AGVs based on time windows according to an embodiment of the present invention. This embodiment proposes a path optimization method for real-time scheduling of multiple AGVs based on time windows, and the method includes the following steps:

[0040] S1, collect the terrain environment information of the unmanned workshop, perform data grid processing, and construct an unmanned workshop model.

[0041] S2. Receive the machine tool material request, allocate an idle trolley, and based on the unmanned workshop model, use the A* algorithm or Dijkstra algorithm to perform initial path planning for the machine tool material request and add corresponding time window markings; the machine tool material request includes a source point and a target node.

[0042] S3. Traverse the routes of the working AGV trolleys one by one, and based on the time window, determine whether there is a time conflict between the unfinished routes and the initial path planned in step S2. If there is no conflict, add this route to the route set P and go to step S6. If there is a conflict, further determine the conflict type and transfer to step S4; the conflict types include concurrent point conflict and head-on conflict.

[0043] S4. According to the conflict type, with the goal of minimizing the number of stop-and-wait times, eliminate the conflict by making one of the AGV trolleys stop and wait, and perform secondary optimization on the initial path.

[0044] S5. Refresh the route set P according to the preset scheduling period, and eliminate the routes that each AGV trolley has already traveled, the route departure points, and the time window markings of the corresponding departure points.

[0045] S6. Repeat steps S2 to S5 until the scheduling of all AGV trolleys is completed.

[0046] I. Construct an unmanned workshop model

[0047] Collect the topographic environment information of the unmanned workshop, perform data grid processing, and construct an unmanned workshop model.

[0048] II. Path planning and adding time window markings

[0049] The system receives the machine tool material request, combines with the unmanned workshop model, and calls the A* algorithm or Dijkstra algorithm according to the machine tool position to obtain a shortest path L a ={r 1 ,r 2 ,r 3 ,...r N}, where N is the number of nodes of this path, and assign this path to an idle AGV trolley a; then according to the departure time t of the trolley, add a time stamp to each departure point of this path to obtain the time stamp set T a ={t 1 ,t 2 ,t 3 ,...t N}, t 1 is the time starting from t and reaching the corresponding first node r 1 , t 2 is the time starting from t and reaching the corresponding second node r 2...t at the time of N is the start time of t and reaches the corresponding Nth node r N at the time. Finally, add this path to the scheduling route set and add the corresponding route timestamp set to the route set P, p = p ∪ {L a , T a}.

[0050] As Figure 2 shown. In this embodiment, an unmanned workshop is taken as an example. There is a material starting center in this workshop, where the parts to be processed are stored; there are 5 public roads for AGV cars to drive; there are 25 machine tools waiting for the processing of parts.

[0051] When the first machine tool sends a request, the system responds to this request. When planning the global static shortest path for each AGV car, considering that the Dijkstra algorithm has a high time complexity and space complexity, while the A* algorithm searches for the shortest path from the source point to the target node, which includes the specific path and is convenient and effective, so this embodiment uses the A* algorithm to implement.

[0052] As Figure 3 , Figure 4 shown, use the A* algorithm to calculate the shortest path and generate a schematic diagram of the shortest path annotation.

[0053] III. Judge whether there is a time conflict and the type of conflict through the time window check method

[0054] First, for the route L a , traverse the departure points r a in L i one by one, where i = 1, 2,..., N, and check whether there is a time conflict with each departure point of each route in P. If the car a detects that the path node is the same as the path node of the car b and the timestamp corresponding to the path node of the car a is the same as the timestamp corresponding to the path node of the car b, it is judged that there is a common point conflict.

[0055] If the car a detects that the path node is the same as the path node of the car b and the next path node of the car a is the same as the previous path node of the car b, and there is an intersection in the timestamps of the corresponding path nodes, it is judged that there is an oncoming conflict.

[0056] If there is no conflict, add this route to the route P and go to step S6.

[0057] IV. Eliminate conflicts according to the conflict type and perform secondary optimization of the route

[0058] If the conflict type belongs to the common point conflict, first mark the path node r of the common point conflict i , at the path node r i+1 Add the path node r i , the original path node r i The subsequent nodes are moved backward by one node position. Pause once at the path node r of the common point conflict i , and then modify the corresponding timestamp set, t i+1 = t i + 1, and modify all the timestamps corresponding to the path nodes after r i+1 by adding 1.

[0059] If the conflict type belongs to the head-on conflict, it is necessary to perform secondary optimization on the head-on conflict path. First, mark the entry path node r of the head-on conflict i , then r i+1 should enter the yield point S. Calculate the time t1 required to reach the yield point S from the entry path node r i , and modify the corresponding timestamp Then, at the yield point S, it is necessary to stop and wait for the AGV in the head-on conflict to drive away from the entry path node r of the head-on conflict i . Calculate the waiting time t2 according to (the distance from the current position of the AGV to the entry path node r i / the speed of the AGV). After that, the AGV needs to return to the entry path node r of the head-on conflict i , modify the route r i+2 = r i , and then modify the corresponding timestamps after by adding t1 + t2, where j ∈ {2, 3, 4,... N - i + 2}.

[0060] As Figure 5 , Figure 6 shown, for example, in this embodiment, there are 2 AGV cars in total, and there is a head-on conflict on their respective travel routes.

[0061] Among them, the coordinates of the No. 1 AGV car are at (1, 14), and its next position is (1, 15); the coordinates of the No. 2 AGV car are at (1, 18), and its next position is (1, 17).

[0062] For the head-on conflict between 2 AGV cars, the specific conflict elimination and path optimization methods are as follows:

[0063] (1) The No. 1 AGV and the No. 2 AGV have an oncoming conflict, and there is an idle obstacle avoidance point to the right of the No. 2 AGV. Therefore, the next position of the No. 2 AGV is updated to (2, 18), and the time from (1, 18) to (2, 18) is calculated to update the time of all subsequent path nodes.

[0064] (2) It is necessary to calculate the waiting time of the No. 2 AGV at the position (1, 17) so that the No. 1 AGV can pass smoothly. The current position of the No. 1 AGV (1, 14) is four grids away from the conflict intersection point (1, 18), and the speed of the AGV is 1 grid per second. Therefore, the No. 2 AGV needs to stay at the obstacle avoidance point (2, 18) for 4 seconds, and then add the original path node path to update the time of all subsequent path nodes.

[0065] The above is only the preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the concept of the present invention belong to the protection scope of the present invention. It should be noted that for those of ordinary skill in the art, several improvements and refinements made without departing from the principle of the present invention should be regarded as within the protection scope of the present invention.

Claims

1. A path optimization method for real-time scheduling of multiple AGVs based on time windows, characterized in that, the method comprises the following steps: S1. Collect the terrain environment information of the unmanned workshop, perform data grid processing, and construct an unmanned workshop model; S2. Receive the machine tool material request, allocate idle trolleys, and combine with the unmanned workshop model. Use the A* algorithm or Dijkstra algorithm to perform initial path planning for the machine tool material request, and add corresponding time window marks; the machine tool material request includes a source point and a target node; S3. Traverse the routes of the working AGV trolleys one by one, and combine with the time window to judge whether there is a time conflict between the unfinished routes and the initial path planned in step S2. If there is no conflict, add the route to the route set P, and go to step S6. If there is a conflict, further judge the conflict type and transfer to step S4; the conflict types include co-point conflict and head-on conflict; S4. According to the conflict type, with the goal of minimizing the number of stop-and-wait times, eliminate the conflict by making one of the AGV trolleys stop and wait, and perform secondary optimization on the initial path; S5. Refresh the route set P according to the preset scheduling period, and eliminate the routes that each AGV trolley has traveled, the route departure points, and the time window marks of the corresponding departure points; S6. Repeat steps S2 to S5 until the scheduling of all AGV trolleys is completed; In step S4, according to the conflict type, with the goal of minimizing the number of stop-and-wait times, eliminate the conflict by making one of the AGV trolleys stop and wait, and the process of performing secondary optimization on the initial path includes the following sub-steps: When the conflict type is a common point conflict, mark the path node r of the common point conflict i , at one of the car's path nodes r i+1 Add path node r i , original path node r i The subsequent nodes are postponed to the next node position, so that the car is at the path node r with common point conflict i Pause once, and then modify the corresponding timestamp set, t i+1 =t i +1, modified r i+1 Then the timestamps corresponding to all path nodes are increased by 1; When the conflict type is head-on conflict, mark the entry path node r of the head-on conflict i , select the path node r of one of the cars i+1 as the yielding point S, and let the car wait at the yielding point S until the head-on conflict car drives away from the entry path node r i ; calculate the time t1 required for the selected car to reach the yielding point S from the entry path node r i , and modify the corresponding timestamp According to the distance from the current position of the head-on conflict car to the entry path node r i , and the speed of the head-on conflict car, calculate the waiting time t2 of the selected car at the yielding point S; when the waiting time ends, the selected car returns to the entry path node r of the head-on conflict i , modify the route r i+2 =r i , and the corresponding timestamps after are all added with t1 + t2, j ∈ {2, 3, 4,... N - i + 2}.

2. The path optimization method for real-time scheduling of multiple AGVs based on time windows according to claim 1, characterized in that, in step S2, the process of performing initial path planning for the machine tool material request and adding corresponding time window marks includes the following sub-steps: S21. Receive the machine tool material request. Combine with the unmanned workshop model. According to the machine tool position, use the A* algorithm or Dijkstra algorithm to obtain an optimal path L a ={r 1 ,r 2 ,r 3 ,...r N}, where N is the number of path nodes. Assign this path to an idle AGV cart a; S22. Add a timestamp to each departure point in path L according to the departure time t of trolley a to obtain the timestamp set T of trolley a a ={t a , t 1 , t 2 ,... t 3 ,... t N}, where t N is the time starting from t and reaching the Nth node r N . S23. Add the scheduling route set and route timestamp set of this path to the route set P, and the specific representation form is as follows: p = p ∪ {L a , T a}。 3. The path optimization method for real-time scheduling of multiple AGVs based on time windows according to claim 2, characterized in that, in step S3, the process of combining with the time window to judge whether there is a time conflict between the unfinished route and the initial path planned in step S2 includes the following sub-steps: Traverse the line L one by one a for the trip point r i where i = 1, 2, …, N, and determine whether there is a time conflict with any trip point of the planned routes in the route set P: If the path nodes of trolley a are detected to be the same as the path nodes of trolley b and the timestamp corresponding to the path nodes of trolley a is the same as the timestamp corresponding to the path nodes of trolley b then it is determined that there is a common point conflict between trolley a and trolley b; If the path nodes of trolley a are detected to be the same as the path nodes of trolley b , and the next path node of trolley a is the same as the previous path node of trolley b , and there is an overlap in the timestamps of the corresponding path nodes, then it is determined that there is a head-on conflict between trolley a and trolley b.

Images