A multi-AGV real-time scheduling and conflict resolution method based on graph structure

Abstract

The present invention discloses a multi-AGV real-time scheduling and conflict resolution method based on a graph structure, including: determining key points on a path according to map data and establishing a graph structure; establishing a global scheduling route; initializing the global scheduling route and an online scheduling route set; assigning the shortest path to an idle AGV car and adding it to the online scheduling route set; the AGV car travels according to the path P; taking the next position of the AGV car from the scheduling route, and performing conflict discrimination with the current position and the next position of other AGV car routes in the online scheduling route set one by one; performing conflict resolution; refreshing the online scheduling route set; repeating the steps until the scheduling is completed. The present invention can effectively resolve conflicts between multiple AGVs, and at the same time can also ensure the scheduling efficiency of multiple AGVs, ensuring that multiple AGVs are collision-free and operate efficiently, and can be fully applied to the real-time scheduling of multiple AGVs in unmanned ports, unmanned workshops or unmanned hatch logistics.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent factories and relates to AGV scheduling technology, and specifically to a multi-AGV real-time scheduling and conflict resolution method based on a graph structure. Background Art

[0002] With the rapid development of computers, manufacturing, warehousing and logistics, and artificial intelligence, a new type of factory that replaces traditional operations has emerged in modern manufacturing, namely, smart factories, also known as unmanned factories or unmanned workshops. Unmanned factories or unmanned workshops, also known as automated factories or workshops, are production factories or workshops that do not require human intervention in production and processing, and can automatically complete tasks such as material distribution, processing, and handling. Because unmanned factories or workshops do not require or only require less human intervention, their production costs will definitely be much lower than those with human participation in a long-term working environment. The emergence of unmanned factories and unmanned workshops has reduced the factory's dependence on people and greatly improved production efficiency. The advantages are particularly obvious when processing precision parts or handling dangerous operations. AGV trolley, the full name of which is "Automated Guided Vehicle", is a flexible intelligent logistics transport trolley that integrates a variety of advanced technologies. The AGV trolley can travel along a specified path and has the ability to automatically load and unload goods. In unmanned factories and unmanned workshops, AGV trolleys can carry materials and transport them to the corresponding machine tools for processing, so AGV trolleys are one of the core equipment for automated production in unmanned factories and unmanned workshops.

[0003] 1. AGV vehicle path planning

[0004] Planning a conflict-free and efficient path for each AGV in an unmanned factory or workshop will be affected by the specific environment of the unmanned factory or workshop. The algorithms for planning paths for AGVs in unmanned factories or workshops are generally Dijkstra algorithm and A* algorithm. AGV path planning mainly includes three aspects:

[0005] (1) Determine whether there is a feasible path from the starting point to the target point.

[0006] (2) The path planned for the AGV must be free of obstruction, conflict, and deadlock.

[0007] (3) The path planned for the AGV should enable the entire system to achieve optimal operating efficiency.

[0008] Laser guidance technology is one of the most important technologies in the development of AGV trolleys. Laser guidance technology is mainly divided into two parts: AGV laser scanner and AGV reflector. The laser scanner installed on the AGV trolley rotates 360° at a fixed speed and sends laser to the reflector. According to the direction of the laser returned by the reflector, the laser scanner can detect the position of the reflector and obtain information about the laser angle. The AGV on-board computer receives and processes the above information to calculate the position and direction of movement of the AGV trolley. During the operation, the AGV trolley sends its current position to the upper control system, which compares and corrects the position of the AGV trolley with the system's built-in parameters about the AGV trolley, thereby guiding the AGV trolley to travel along the correct route. In addition, in the laser guidance system, a certain number of reflectors are generally strategically installed in the designated area. This is to enable the AGV to feedback the current position of the AGV trolley from multiple reflectors at different angles when traveling in the designated area, so as to improve the accuracy of the AGV trolley's position.

[0009] 2. Multi-AGV Scheduling Method

[0010] The scheduling problem of multiple AGVs refers to using certain parameters as optimization targets, determining a reasonable resource allocation plan based on tasks and specific factory resources, allocating the material transportation tasks of the unmanned workshop to each AGV, and scheduling the AGVs for collision-free operation, so as to complete the transportation tasks with a higher efficiency.

[0011] At present, the scheduling methods that are widely used and studied include scheduling algorithms based on time window path planning, scheduling algorithms based on conflict avoidance based on graph or grid structure, and scheduling algorithms based on artificial intelligence prediction.

[0012] The scheduling algorithm based on time window path planning is a simple and practical scheduling method that can avoid conflicts and waiting deadlocks in AGV workshops. The method stipulates that in a certain time period, a certain road section is exclusively occupied by a certain AGV car, and other AGV cars are not allowed to enter the road section in this time period. The conflict types based on time windows are generally divided into common point conflict types and opposite direction conflict types. (1) Common point conflict: If in a certain time period, an AGV car enters a position exclusively occupied by another AGV car, then the AGV car waits and increases the timestamp in the path by one. (2) Opposite direction conflict: If in a certain time period, an AGV car and another AGV car are driving in opposite directions and are about to conflict, then the AGV car drives to a yield point to give way to the other AGV car. For example, "Multi-AGV Scheduling Method" (Patent No.: 201610999699.6) proposes a scheduling method based on a topological map, which performs conflict judgment and path planning based on time windows. Scheduling based on time windows relies on the calculation of absolute time, which is inconsistent with the actual operating environment. For example, time calculation errors caused by vehicle failure or deceleration may lead to scheduling failure.

[0013] The conflict avoidance scheduling algorithm based on graph or grid structure formulates corresponding conflict resolution strategies according to the different conflict types when multiple AGVs are running, so as to avoid conflicts when AGVs are running and improve the operation efficiency of multiple AGVs. Common conflict types include opposite conflict types, intersection conflict types and node occupancy conflict types. For example, "A multi-AGV scheduling method based on dynamic node detection" (publication number: CN108364100) adopts a graph-based stop-and-wait strategy for conflict resolution, while the stop-and-wait strategy is not suitable for all situations. For example, in the case of opposite conflicts, the stop-and-wait strategy may cause long waiting or even deadlock. "A grid map and a multi-AGV scheduling method based on a grid map" (CN109557886) adopts a grid map and proposes a scheduling method for path search based on idle marks, busy marks and obstacle marks. The scheduling process of this method is computationally intensive and the path is not optimal.

[0014] Traditional conflict avoidance strategies include re-planning the path, stopping and waiting, etc. When adopting conflict avoidance methods, it is necessary to select the appropriate conflict resolution method according to the actual environment of the unmanned workshop. Re-planning the path may reduce the efficiency of conflict avoidance and cause a certain waste of resources; therefore, different obstacle avoidance strategies should be adopted for different types of conflicts. Summary of the invention

[0015] Purpose of the invention: In order to overcome the deficiencies in the prior art, a multi-AGV real-time scheduling and conflict resolution method based on a graph structure is provided, which can effectively resolve the conflicts between multiple AGVs while also ensuring the scheduling efficiency of multiple AGVs, ensuring that multiple AGVs operate without collision and with high efficiency, and can be fully applied to the real-time scheduling of multiple AGVs in unmanned ports, unmanned workshops or unmanned hatch logistics.

[0016] Technical solution: To achieve the above purpose, the present invention provides a multi-AGV real-time scheduling and conflict resolution method based on a graph structure, comprising the following steps:

[0017] S1: Determine the key points on the path based on the map data and establish the graph structure;

[0018] S2: Establish a global scheduling route based on the graph structure;

[0019] S3: Start scheduling. First, read in the initialized global scheduling route and initialize the online scheduling route set R to be empty.

[0020] S4: According to the AGV request instruction and the global scheduling route, the shortest path P is assigned to the idle AGV, and the path P is added to the online scheduling path set R, that is, R = RυP;

[0021] S5: The AGV travels along the path P. When an AGV arrives at the next designated position and requests instructions, if the position is the end point and the task has been completed, the task is terminated, the AGV is released and returns to the designated position, and is in an idle waiting state. If the next position is not the end point, go to step S6 for conflict judgment;

[0022] S6: Take the next position of the AGV from the scheduling route, and perform conflict determination with the current position and next position of other AGV routes in the online scheduling route set R one by one. If there is no conflict with all the next position points, the next position is sent to the AGV and executed, and the process goes to step S8. If there is a conflict with other AGV routes, the process goes to step S7.

[0023] S7: resolve the conflict according to the conflict type and give instructions to get off the vehicle;

[0024] S8: In a scheduling cycle, the system refreshes the online scheduling route set R to eliminate the routes that have been traveled and the remaining travel points of the routes, as well as the remaining distance from the current travel to the next location;

[0025] S9: Repeat steps S5 to S8 until the scheduling is completed.

[0026] Furthermore, the conflict determination in step S6 includes a conflict determination of common points but not common lines and a conflict determination of common points and common lines.

[0027] Furthermore, the specific method for determining the common point and non-collinear conflict in step S6 is as follows:

[0028] If there is a common point conflict and they are not collinear, it is necessary to calculate the time it takes for each vehicle to arrive at the next position to determine whether they arrive at the same time. The calculation method is: the remaining distance D to the next position point divided by the speed V of each vehicle, that is, t=D / V. If their time difference is greater than the set threshold T, there is no conflict; if the time difference is less than T, there is a conflict, and then go to step S7 for conflict resolution.

[0029] Furthermore, the specific method of determining the common point and common line conflict in step S6 is: if there is a common point conflict and a common line, then there is a conflict, and then the process goes to step S7 to perform conflict resolution processing.

[0030] Furthermore, the threshold T is determined by:

[0031]

[0032] Among them, L is the length of the AGV body, W is the width of the AGV, and V is the driving speed of the AGV.

[0033] Furthermore, step S7 includes conflict resolution for co-point conflicts but not co-linear conflicts and conflict resolution for co-point and co-linear conflicts.

[0034] The specific method for resolving conflicts that are co-point conflicts and not co-linear conflicts is:

[0035] If there is a common point conflict and they are not on the same line, and they arrive at the next position at the same time, then the car with the lower priority is assigned a pause time K, and another car is assigned to the next position and executed.

[0036] The formula for determining the time K is:

[0037] K=1.5×T

[0038] Among them, L is the length of the AGV car body, W is the width of the AGV car, and V is the driving speed of the AGV car. The specific method for resolving the conflict of co-points and co-lines is:

[0039] If they share the same point and line, there is a conflict between them. The car with lower priority is designated to avoid it, and an avoidance route is planned for it. The avoidance route is added to the current route and executed. The starting point and end point of the avoidance route are both the current give-way point, and it must pass through the avoidance point. The avoidance point is determined by the avoidance position closest to the current give-way point.

[0040] The present invention models the specific environment of the unmanned workshop with a graph structure, and then uses a path planning algorithm to plan a global shortest path for each pair of machine tools and unloading points. When the machine tool sends a parts request to the material starting center, the dispatch center selects an idle AGV car from the idle AGV car queue, and gives the global shortest path of the corresponding machine tool to the AGV car. The AGV car drives according to the planned global shortest path by default. When a conflict occurs, a conflict resolution method based on a graph structure is used to determine whether it is a common point conflict or a opposite conflict, and different resolution methods are used for different conflict types.

[0041] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0042] 1. The method proposed in the present invention can be applied to most unmanned workshops and logistics transportation environments and is universal. It is only necessary to grid the data of the unmanned workshop and establish a graph structure model, and then the method proposed in the present invention can be introduced to model the unmanned workshop map.

[0043] 2. The multi-AGV conflict resolution method based on graph structure proposed in the present invention is more efficient than the traditional direct common point conflict judgment and resolution method, ensuring that multiple AGVs can operate without collision and with high efficiency.

[0044] 3. After the present invention establishes the map of the unmanned workshop, it can easily simulate the scheduling of multiple AGVs, and can clearly observe the entire process of AGV conflict resolution, verifying the correctness and high efficiency of the method provided by the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a general flow chart of the method of the present invention;

[0046] Figure 2 Modeling maps for port logistics;

[0047] Figure 3 It is the path planning graph of A* algorithm;

[0048] Figure 4 Implement the corresponding map annotation diagram for the A* algorithm;

[0049] Figure 5 The first conflict case Figure 1 ;

[0050] Figure 6 The first conflict case Figure 2 ;

[0051] Figure 7 The first conflict case Figure 3 ;

[0052] Figure 8 For the second conflict case Figure 1 ;

[0053] Fig. 9 For the second conflict case Figure 2 ;

[0054] Fig.10 For the second conflict case Figure 3 ;

[0055] Fig.11 The third conflict case Figure 1 ;

[0056] Fig.12 The third conflict case Figure 2 ;

[0057] Fig.13 The third conflict case Figure 3 ;

[0058] Fig.14 This is a multi-AGV simulation operation diagram. DETAILED DESCRIPTION

[0059] The present invention is further explained below in conjunction with the accompanying drawings and specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, various equivalent forms of modifications to the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0060] The present invention provides a multi-AGV real-time scheduling and conflict resolution method based on a graph structure, which is applied to unmanned workshops. Figure 1 , which includes the following steps 1 to 7:

[0061] Step 1: Data preparation and initialization

[0062] Prepare the map data of the unmanned workshop, determine the key points of the route, including route intersections, loading points, unloading points (machine tool locations), avoidance points (can be specific avoidance positions for a single line, can be double-line avoidance positions, or fall back to a certain position without conflict, etc.), measure the distance between points, and then establish a graph structure model. According to the loading points and unloading points, a global route is established with the help of path planning algorithms such as A* or Dijstra. At the beginning of scheduling, the system reads the map data and global route of the unmanned workshop, and initializes the online scheduling route set R to be empty.

[0063] Step 2: Assign tasks

[0064] Process the delivery request of each machine tool one by one. The system responds to the request of a machine tool, searches for a global shortest path from the global path according to the location of the machine tool, assigns this path to an idle AGV, and adds the path to the online scheduling path set.

[0065] Step 3: Transport Scheduling:

[0066] Process the AGV's request for the next position on the route one by one. When an AGV arrives at the designated next position and requests instructions, if the position is the end point (unloading point) and the task has been completed, the task is ended, the AGV is released and returns to the designated position, in an idle waiting state. If the next position is not the end point, go to step 4 for conflict judgment.

[0067] Step 4: Conflict Identification

[0068] The next position of the AGV is taken from the scheduling route, and conflicts are determined one by one with the next positions of other AGV routes in the online scheduling route set R. The specific determination method is as follows:

[0069] (1) Conflict judgment for co-points but not co-linearities. If there is a co-point conflict and they are not co-linear, it is necessary to calculate the time it takes for each vehicle to reach the next location to determine whether they arrive at the same time. The calculation method is: the remaining distance D to the next location point divided by the speed V of each vehicle, that is, t = D / V. If their time difference is greater than a certain threshold T, there is no conflict; if the time difference is less than T, there is a conflict, and go to step 5 for conflict resolution. T is determined based on the vehicle body length L, width W and its driving speed V:

[0070]

[0071] (2) Conflict judgment of common points and common lines. If there is a conflict of common points and common lines, there is a conflict, and then go to step 5 for conflict handling.

[0072] If there is no conflict with all the next position points, the next position is sent to the AGV and executed, and then go to step 6.

[0073] Step 5: Route conflict resolution

[0074] Conflict resolution is performed based on the conflict type:

[0075] (1) If there is a common point conflict and they are not collinear, and they arrive at the next position at the same time, then the car with the lower priority is designated to pause for a period of time K, and another car is designated to the next position. Otherwise, the cars are instructed to continue to drive to the next position. K is determined based on the car body length L, width W and driving speed V, which is equivalent to ensuring that the time interval is greater than 3 times the time of driving one car body length plus the car width to avoid collision. The calculation formula is as follows:

[0076] K=1.5×T

[0077] (2) If there are common points and common lines, there is a conflict. The car with lower priority is designated to avoid, and a avoidance route is planned for it. The avoidance route is added to the current route and the next position is indicated. The starting point and end point of the avoidance route are both the current give-way point, and it must pass through the avoidance point. The avoidance point is determined by the avoidance position closest to the current give-way point.

[0078] Step 6: Online route update

[0079] Refresh the online scheduling route set R. In a scheduling cycle, the system refreshes the route set R to update the distance from the current position to the next position of each car, as well as the travel points of the route being executed.

[0080] Step 7: Repeat steps 3 to 6 until the scheduling is completed.

[0081] Based on the above content, in order to verify the actual effect of the method of the present invention, the solution of the present invention is applied to an unmanned port, specifically, the solution of the present invention is applied as follows:

[0082] After importing the data of the unmanned port, the graph model of the unmanned port was applied, such as Figure 2 As shown. Figure 2 Take the unmanned port in China as an example. There is a container ship and four quay cranes in the port, which store several containers waiting to be taken out; there are 81 roads for AGV vehicles to drive; and there are 5 yards waiting for the recovery of containers.

[0083] In this embodiment, when planning the global static shortest path for each AGV, the Dijkstra algorithm has high time complexity and space complexity, while the A* algorithm searches for the shortest path from the source node to the target node, which includes the specific path and is convenient and effective, so the A* algorithm is used for implementation. Figure 2 In the map environment of an unmanned port, the A* algorithm is used to calculate the shortest path, such as Figure 3 As shown, the corresponding shortest path is marked as Figure 4 shown.

[0084] The conflict resolution method proposed in this embodiment can effectively resolve conflicts between multiple AGVs. The specific conflict resolution scheme is as described in the implementation steps above. In order to facilitate the observation of accurate graph information, part of the operation graph is intercepted.

[0085] (1) Figure 5-6 -7 is the first case diagram when a conflict occurs. There are 3 AGVs in the case, and a conflict occurs at point 46. Among them, the orange AGV No. 1 is located at point 32, and its next position is point 46; the blue AGV No. 3 is located at point 45, and its next position is point 46; the purple AGV No. 4 is located at point 6, and its next position is point 20.

[0086] Specific conflict resolution methods:

[0087] The orange AGV No. 1 and the blue AGV No. 3 have a point conflict at point 46. Since the blue AGV No. 3 has a higher priority, if they arrive at the same time, the blue AGV No. 3 will be given priority to be assigned the next route, so the blue AGV No. 3 will go first. The orange AGV No. 1 stops and waits.

[0088] (2) Figure 8-9 -10 is the second case diagram when a conflict occurs. There are 3 AGVs in the case, and a conflict occurs at point 46. Among them, the orange AGV No. 1 is located between points 33 and 46, and its next position is point 46; the blue AGV No. 3 is located at point 45, and its next position is point 46; the purple AGV No. 4 is located between points 35 and 36, and its next position is point 36.

[0089] Specific conflict resolution methods:

[0090] The orange AGV No. 1 and the blue AGV No. 3 have a co-point conflict at point 46. Since the orange AGV No. 1 is on the route 33-46 first, according to the real-time scheduling update algorithm, the estimated arrival time of point 46 of the orange AGV No. 1 is greater than the estimated arrival time of point 46 of the blue AGV No. 3, so the scheduling system will still assign the next route to the blue AGV No. 3.

[0091] (3) Figure 11-12 -13 is the third case diagram when a conflict occurs. There are 2 AGVs in the case, and a conflict occurs at point 46. Among them, the orange AGV No. 1 is located between points 33 and 46, and its next position is point 46; the blue AGV No. 3 is located at point 45, and its next position is point 46.

[0092] Specific conflict resolution methods:

[0093] The orange AGV No. 1 and the blue AGV No. 3 have a co-point conflict at point 46. Since the orange AGV No. 1 is on the route 33-46 first, according to the real-time scheduling update algorithm, the estimated arrival time of the orange AGV No. 1 at point 46 is less than the estimated arrival time of the blue AGV No. 3 at point 46, so the scheduling system will let the blue AGV No. 3 stop and wait.

[0094] In this embodiment, the graph-based data of the unmanned port is imported to build a map of the unmanned port and facilitate the simulation of multiple AGVs. Figure 2 Based on the established map model of the unmanned port, the dispatching process of multiple AGVs is simulated, so that users can easily observe the dispatching process of multiple AGVs. Fig.14 As shown, the effectiveness and correctness of the conflict resolution method proposed in the present invention are verified.

Claims

1. A multi-AGV real-time scheduling and conflict resolution method based on graph structure, characterized in that: The steps include: S1: Determine the key points on the path based on the map data and establish the graph structure; S2: Establish a global scheduling route based on the graph structure; S3: Start scheduling. First, read in the initialized global scheduling route and initialize the online scheduling route set R to be empty. S4: According to the AGV request instruction and the global scheduling route, the shortest path P is assigned to the idle AGV, and the path P is added to the online scheduling path set R, that is, R = RυP; S5: The AGV travels along the path P. When an AGV arrives at the next designated position and requests instructions, if the position is the end point and the task has been completed, the task is terminated, the AGV is released and returns to the designated position, and is in an idle waiting state. If the next position is not the end point, go to step S6 for conflict judgment; S6: Take the next position of the AGV from the scheduling route, and perform conflict determination with the current position and next position of other AGV routes in the online scheduling route set R one by one. If there is no conflict with all the next position points, the next position is sent to the AGV and executed, and the process goes to step S8. If there is a conflict with other AGV routes, the process goes to step S7. S7: resolve the conflict according to the conflict type and give instructions to get off the vehicle; S8: In a scheduling cycle, the system refreshes the online scheduling route set R to eliminate the routes that have been traveled and the remaining travel points of the routes, as well as the remaining distance from the current travel to the next location; S9: Repeat steps S5 to S8 until the scheduling is completed.

2. According to the graph-structured multi-AGV real-time scheduling and conflict resolution method of claim 1, it is characterized in that: The conflict determination in step S6 includes the conflict determination of common points but not common lines and the conflict determination of common points and common lines.

3. The method for real-time scheduling and conflict resolution of multiple AGVs based on a graph structure according to claim 2 is characterized in that: The specific method for determining the common point and non-collinear conflict in step S6 is: If there is a common point conflict and they are not collinear, it is necessary to calculate the time it takes for each vehicle to arrive at the next position to determine whether they arrive at the same time. The calculation method is: the remaining distance D to the next position point divided by the speed V of each vehicle, that is, t=D / V. If their time difference is greater than the set threshold T, there is no conflict; if the time difference is less than T, there is a conflict, and then go to step S7 for conflict resolution.

4. The method for real-time scheduling and conflict resolution of multiple AGVs based on a graph structure according to claim 2 is characterized in that: The specific method of determining the common point and common line conflict in step S6 is: if there is a common point conflict and a common line, then there is a conflict, and then the process goes to step S7 to perform conflict resolution processing.

5. The method for real-time scheduling and conflict resolution of multiple AGVs based on graph structure according to claim 3 is characterized in that: The threshold value T is determined by: Among them, L is the length of the AGV body, W is the width of the AGV, and V is the driving speed of the AGV.

6. The method for real-time scheduling and conflict resolution of multiple AGVs based on graph structure according to claim 2 is characterized in that: The step S7 includes conflict resolution for co-point conflicts but not co-linear conflicts and conflict resolution for co-point and co-linear conflicts.

7. The method for real-time scheduling and conflict resolution of multiple AGVs based on graph structure according to claim 6 is characterized in that: The specific method for resolving the conflict of co-points but not co-lines in step S7 is: If there is a common point conflict and they are not on the same line, and they arrive at the next position at the same time, then the car with the lower priority is assigned a pause time K, and another car is assigned to the next position and executed.

8. The method for real-time scheduling and conflict resolution of multiple AGVs based on graph structure according to claim 6 is characterized in that: The specific method for resolving the conflict of co-points and co-lines in step S7 is: If they share the same point and line, there is a conflict between them. The car with lower priority is designated to avoid it, and an avoidance route is planned for it. The avoidance route is added to the current route and executed. The starting point and end point of the avoidance route are both the current give-way point, and it must pass through the avoidance point. The avoidance point is determined by the avoidance position closest to the current give-way point.

9. The method for real-time scheduling and conflict resolution of multiple AGVs based on graph structure according to claim 7 is characterized in that: The formula for determining the time K is: Among them, L is the length of the AGV body, W is the width of the AGV, and V is the driving speed of the AGV.

Images