Multi-AGV intersection coordination method for automated container terminals

Abstract

The present invention relates to the field of path planning technology, and specifically discloses a multi-AGV intersection coordination method for an automated container terminal, comprising: determining the current number of intersections and the number of AGVs, and constructing an intersection traffic model according to the current number of intersections and the number of AGVs, wherein the intersection traffic model at least includes a polygonal intersection model; searching the intersection traffic model for a traffic order according to an MCTS algorithm to obtain the traffic priority of each AGV; sending the traffic priority of each AGV to the corresponding AGV, and each AGV can perform transportation operations according to its own traffic priority. The multi-AGV intersection coordination method for an automated container terminal provided by the present invention realizes conflict-free deadlock and efficient multi-AGV intersection coordinated traffic.

Description

Technical Field

[0001] The present invention relates to the technical field of path planning, and in particular to a multi-AGV intersection coordination method for an automated container terminal. Background Art

[0002] In the modern logistics industry, ocean transportation is the main mode of transportation in the international logistics industry. Ocean transportation has become the pillar of international trade and the key driving force for the globalization of trade. As an emerging transportation equipment, the Automatic Guided Vehicle (AGV) can complete the container transportation and loading and unloading operations between port loading and unloading ships and yards. It is equipped with a variety of sensors to achieve positioning and navigation functions, which is convenient for system scheduling operations and meets operational needs in a faster and more convenient way.

[0003] Due to the complex road network structure and large number of tasks, automated terminals require a large number of AGVs to operate simultaneously, which can lead to intersection conflicts and deadlocks during the operation of AGVs. Especially at intersections with complex driving routes, if the AGV's intersection traffic plan is not planned through the intersection collaborative traffic method, a large number of intersection conflicts and deadlocks will inevitably occur, thus affecting the safety and efficiency of port operations.

[0004] Most existing AGV intersection coordination solutions have simplified the intersections of automated container terminals, which is quite different from the actual port operation environment. Some of the solutions simplify the intersection area into a node, without considering the geometric properties of the intersection, resulting in only one AGV passing through the intersection area at the same time. On this basis, node time windows and other methods are used to plan the passage plan of AGVs at the intersection. The node time window is divided based on the predicted arrival time of each AGV. In the actual operation process, the congestion between AGVs will increase the prediction error, resulting in poor efficiency of this method. Another part of the solution grids the intersection area, and each grid will dynamically maintain an AGV occupancy sequence in real time. Based on this sequence, conflicts between AGVs are detected to achieve coordinated passage of multiple AGVs at the intersection. The efficiency of the grid method is related to the density of the grid division. The finer the grid division, the more effective the grid method is, but this will also lead to a significant increase in its computational workload, thereby affecting the computational efficiency of the system in actual operation.

[0005] In urban scenarios, motion control and trajectory prediction are used to achieve intersection coordination. However, due to the open urban environment, the connection between intersections is not close. Most solutions only focus on a single intersection and do not consider the mutual influence of the results of intersection coordination methods between intersections. Automated container terminals are relatively closed, the connection between intersections is close, and there are a large number of AGVs performing transportation operations. The impact of the results of intersection coordination methods between intersections will also be more important.

[0006] Therefore, how to provide a solution for the coordinated passage of multiple AGVs at intersections has become a technical problem that needs to be solved urgently by those skilled in the art. Summary of the invention

[0007] The present invention provides a multi-AGV intersection coordination method for an automated container terminal, which solves the problem of low traffic efficiency caused by lack of intersection coordination when AGVs perform transportation operations in the related art.

[0008] As one aspect of the present invention, a method for coordinating multiple AGV intersections at an automated container terminal is provided, which includes:

[0009] Determine the current number of intersections and the number of AGVs, and construct an intersection traffic model according to the current number of intersections and the number of AGVs, wherein the intersection traffic model at least includes a polygonal intersection model;

[0010] According to the MCTS algorithm, the traffic order of the intersection traffic model is searched to obtain the traffic priority of each AGV;

[0011] The traffic priority of each AGV is sent to the corresponding AGV, and each AGV can perform transportation operations according to its own traffic priority;

[0012] The traffic order of the intersection traffic model is searched according to the MCTS algorithm to obtain the traffic priority of each AGV, including:

[0013] According to the MCTS algorithm, the optimization goal of the intersection traffic model is to minimize the average AGV intersection passing time;

[0014] For each intersection, the future path table of each AGV is filtered to obtain the combined passage mode of AGVs without conflicts.

[0015] Find the best traffic sequence for each intersection in the AGV combined traffic mode according to the MCTS algorithm and the optimization target of the traffic model at the intersection;

[0016] Determine the passage priority of each AGV at each intersection according to the optimal passage order;

[0017] According to the MCTS algorithm and the optimization target of the intersection traffic model, the optimal traffic sequence of each intersection in the AGV combined traffic mode is found, including:

[0018] A search tree is constructed for each intersection, wherein the search tree includes a root node and multiple layers of nodes extending sequentially from the root node, each node includes a combination of AGVs that can pass through at the same time, and each node can expand downward to a child node, the node at the first layer has the highest priority, and each branch of the search tree starting from the root node represents a feasible passage order;

[0019] Iterate the search tree according to the MCTS algorithm until the optimization goal of the intersection traffic model is met, and obtain the optimal traffic sequence of each intersection in the AGV combined traffic mode;

[0020] Iterating the search tree according to the MCTS algorithm includes:

[0021] Starting from the root node, a node to be expanded is selected by a simulation strategy formula, wherein the simulation strategy formula is expressed as:

[0022] ,

[0023] in, Represents a child node The score, Indicates the number of times the current node has been visited. Represents a child node The number of times it has been visited. represents the weight coefficient;

[0024] According to the nodes selected by the simulation strategy formula, new nodes are gradually expanded downward to reach the target state, wherein the target state is a state that can include all elements of the AGV set;

[0025] Calculate the score of the current node;

[0026] Backpropagate based on the score of the current node to update the score of the parent node.

[0027] Furthermore, the current number of intersections and the number of AGVs are determined, and an intersection traffic model is constructed based on the current number of intersections and the number of AGVs, including:

[0028] Get all current intersection information and all current AGV information;

[0029] Generate an intersection list according to all current intersection information, and generate an AGV list and an AGV future path list according to all current AGV information, wherein each AGV future path list includes multiple waypoint information;

[0030] Determine the set of AGVs that are waiting to pass through at any intersection and at the same time for each AGV based on the AGV future path list;

[0031] A parking area model for an intersection is constructed, and a traffic priority set for the AGV set is constructed.

[0032] Furthermore, a parking area model at the intersection is constructed, including:

[0033] Construct a polygonal model matching the shape of the intersection according to the shape of the intersection;

[0034] A peripheral control line is arranged at a preset distance from the polygonal area on the periphery of the polygonal model, and a control area is formed between the peripheral control line and the polygonal model;

[0035] When each AGV enters the control area of ​​the current intersection, the AGV is added to the AGV set, and when the AGV leaves the control area of ​​the current intersection, the AGV is deleted from the AGV set.

[0036] Furthermore, a traffic priority set of the AGV set is constructed, including:

[0037] Determine the real-time assigned traffic priority of each AGV in the AGV set;

[0038] A traffic priority set of the AGV set is determined according to the real-time assigned traffic priority of each AGV.

[0039] Furthermore, for each intersection, the combined passage mode of AGVs without conflicts is obtained after screening according to the future path table of each AGV, including:

[0040] For each intersection, determine whether there is a conflict in the passage of AGVs based on the future path table of each AGV;

[0041] If it is determined that there is an AGV combination with a passage conflict, the AGV combination with the passage conflict is removed, and a passage mode is obtained by removing the AGV combination with the passage conflict.

[0042] Furthermore, judging whether there is a conflict in the passage of the AGV according to the future path table of each AGV includes:

[0043] For two AGVs, determine the future path sets corresponding to each AGV as R1 and R2 respectively;

[0044] The minimum distance between the future paths of two AGVs is calculated according to the minimum distance calculation formula, where the minimum distance calculation formula is:

[0045] ,

[0046] in, represents the minimum distance between all elements in R1 and all elements in R2, Indicates the minimum safe distance between two AGVs;

[0047] If the minimum distance formula is satisfied, it is determined that there is no conflict between the two AGVs;

[0048] If the minimum distance formula is not satisfied, it is determined that there is a conflict between the two AGVs.

[0049] Furthermore, the score of the current node is calculated, including:

[0050] The first part of the score is determined according to the branch layer number where the current node is located. The calculation formula of the first part of the score is:

[0051] ,

[0052] in, represents the first part of the score, Indicates the level of the branch where the current node is located. represents the corresponding weight coefficient;

[0053] The second part of the score is determined according to the number of AGVs included in the current node and the number of AGVs queued behind it. The calculation formula of the second part of the score is:

[0054] ,

[0055] in, represents the second part of the score, Indicates the AGV set included in the current node Middle The number of AGVs that queue up after each element, represents the corresponding weight coefficient;

[0056] The third part score is determined according to the collaborative impact results of adjacent intersections. The calculation formula of the third part score is:

[0057] ,

[0058] in, represents the third part of the score, Indicates the current intersection. express The adjacent intersections, Indicates that you are currently in The next intersection is And through The lane and The same total number of AGVs, Indicates the AGV set included in the current node Middle AGVs, Indicates the distance between the adjacent intersection and the next intersection. represents the corresponding weight coefficient;

[0059] The score of the current node is calculated based on the first part score, the second part score and the third part score. The calculation formula is:

[0060] .

[0061] The automated container terminal multi-AGV intersection coordination method provided by the present invention determines the current number of intersections and the number of AGVs, constructs an intersection traffic model, and then searches the traffic order of the intersection traffic model according to the MCTS algorithm to obtain the AGV traffic priority, and sends the AGV traffic priority to the corresponding AGV. Therefore, the automated container terminal multi-AGV intersection coordination method of the present invention periodically assigns traffic priority to the AGVs waiting to pass through the intersection, and after considering the three aspects of simultaneous passage of multiple AGVs, queue relief, and congestion avoidance, comprehensively improves and applies MCTS to search for the best traffic order, and realizes conflict-free deadlock and efficient multi-AGV intersection coordinated passage. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present invention, but do not constitute a limitation of the present invention.

[0063] Figure 1 This is a flow chart of the multi-AGV intersection coordination method for an automated container terminal provided by the present invention.

[0064] Figure 2 A flow chart of constructing an intersection traffic model provided by the present invention.

[0065] Figure 3 A flow chart of constructing an intersection parking area model provided by the present invention.

[0066] Figure 4 A flow chart for obtaining the passage priority of each AGV provided by the present invention.

[0067] Figure 5 A flowchart for finding the optimal passage sequence according to the MCTS algorithm provided by the present invention.

[0068] Figure 6A schematic diagram of the basic process of constructing a search tree for MCTS provided by the present invention.

[0069] Figure 7 A schematic diagram of the simultaneous passage of multiple AGVs and their corresponding search trees provided by the present invention.

[0070] Figure 8 This is a schematic diagram of AGV queue relief provided by the present invention.

[0071] Fig. 9 A schematic diagram of the impact of adjacent intersection coordination between intersections provided by the present invention. DETAILED DESCRIPTION

[0072] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0073] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment 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 making creative work should fall within the scope of protection of the present invention.

[0074] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so as to describe the embodiments of the present invention described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0075] In this embodiment, a multi-AGV intersection coordination method for an automated container terminal is provided. Figure 1 is a flow chart of a multi-AGV intersection coordination method for an automated container terminal provided according to an embodiment of the present invention. Figure 1 As shown, including:

[0076] S100, determining the current number of intersections and the number of AGVs, and constructing an intersection traffic model according to the current number of intersections and the number of AGVs, wherein the intersection traffic model at least includes a polygonal intersection model;

[0077] In the embodiment of the present invention, all the intersections and AGVs are obtained, the data structure required for the subsequent MCTS (Monte Carlo Tree Search) algorithm is constructed, and the model is built and various parameters are initialized.

[0078] S200, searching the traffic order of the intersection traffic model according to the MCTS algorithm to obtain the traffic priority of each AGV;

[0079] In an embodiment of the present invention, the MCTS algorithm is introduced to comprehensively consider the simultaneous passage of multiple AGVs, queue relief and congestion avoidance, and the MCTS algorithm is improved. By periodically applying this method between adjacent intersections, the passage order is planned for the AGVs waiting to pass through the intersection.

[0080] S300: Send the traffic priority of each AGV to the corresponding AGV, and each AGV can perform transportation operations according to its own traffic priority.

[0081] It should be understood that when the AGV arrives at the intersection, the MCTS algorithm can assign it a traffic priority.

[0082] Therefore, the automated container terminal multi-AGV intersection coordination method provided by the present invention determines the current number of intersections and the number of AGVs, constructs an intersection traffic model, and then searches the traffic order of the intersection traffic model according to the MCTS algorithm to obtain the AGV traffic priority, and sends the AGV traffic priority to the corresponding AGV. Therefore, the automated container terminal multi-AGV intersection coordination method of the present invention periodically assigns traffic priority to AGVs waiting to pass through the intersection, and after considering the three aspects of simultaneous passage of multiple AGVs, queue relief, and congestion avoidance, comprehensively improves and applies MCTS to search for the best traffic order, and realizes conflict-free deadlock and efficient multi-AGV intersection coordinated passage.

[0083] In the embodiment of the present invention, the current number of intersections and the number of AGVs are determined, and an intersection traffic model is constructed according to the current number of intersections and the number of AGVs, such as Figure 2 As shown, including:

[0084] S110, obtaining all current intersection information and all current AGV information;

[0085] In this embodiment of the present invention, a list of all AGVs is obtained. , intersection list , List of future paths for AGV ,The future path of each AGV consists of a series of waypoints.

[0086] S120, generating an intersection list according to all current intersection information, and generating an AGV list and an AGV future path list according to all current AGV information, wherein each AGV future path list includes a plurality of waypoint information;

[0087] The specific intersection list information is as follows:

[0088] ,

[0089] ,

[0090] ,

[0091] in, Indicates AGVs, for The future path of All by coordinates express; Indicates intersections; and Represent the number of AGVs and the number of intersections respectively.

[0092] S130, determining, according to the AGV future path list, a set of AGVs that each AGV is waiting to pass through at any intersection and at the same time;

[0093] It should be understood that for AGV , it will pass through multiple intersections during the operation, and the collection of these intersections is used Indicates; for intersection , there will be multiple AGVs waiting to pass at the same time , the collection of these AGVs is used express.

[0094] ,

[0095] ,

[0096] in, Representing a collection The number of elements in ; Representing a collection The number of elements.

[0097] S140: constructing a parking area model for an intersection, and constructing a set of traffic priorities for the set of AGVs.

[0098] The intersection area is actually a polygon. When each AGV is about to reach the intersection, it should slow down and stop outside the intersection area.

[0099] In the embodiment of the present invention, a parking area model of an intersection is constructed, such as Figure 3 As shown, including:

[0100] S141, constructing a polygonal model matching the shape of the intersection according to the shape of the intersection;

[0101] S142, setting a peripheral control line at a preset distance from the polygonal area on the periphery of the polygonal model, so that a control area is formed between the peripheral control line and the polygonal model;

[0102] When each AGV enters the control area of ​​the current intersection, the AGV is added to the AGV set, and when the AGV leaves the control area of ​​the current intersection, the AGV is deleted from the AGV set.

[0103] It should be noted that the embodiment of the present invention sets a control zone by expanding the boundary of the intersection area outward by a certain distance to provide a buffer for the AGV to slow down and stop. The corresponding control zone is added to the set by the algorithm. When it leaves the intersection When Remove from.

[0104] In an embodiment of the present invention, constructing a set of traffic priorities of the AGV set includes:

[0105] Determine the real-time assigned traffic priority of each AGV in the AGV set;

[0106] A traffic priority set of the AGV set is determined according to the real-time assigned traffic priority of each AGV.

[0107] It should be understood that at regular intervals, the MCTS algorithm is The AGVs in the game are assigned traffic priorities, using express The set of access priorities of all AGVs in .

[0108] ,

[0109] The optimization goal of the MCTS algorithm is to minimize the average AGV intersection passing time, that is:

[0110] ,

[0111] in, yes pass The time spent express The elements.

[0112] In the embodiment of the present invention, the traffic order search is performed on the intersection traffic model according to the MCTS algorithm to obtain the traffic priority of each AGV, such as Figure 4 As shown, including:

[0113] S210, determining, according to the MCTS algorithm, that the optimization goal of the intersection traffic model is to minimize the average AGV intersection passing time;

[0114] S220, for each intersection, obtain a combined passage mode of AGVs without conflicts after screening according to the future path table of each AGV;

[0115] For intersections ,That In the collection AGVs, without considering the path conflicts between AGVs, the number of combined passing modes for:

[0116] .

[0117] In order to reduce computational redundancy, we first determine whether there is a conflict between AGVs based on the future path of the AGV, and then screen out the conflicting AGV combination traffic modes.

[0118] Specifically, for each intersection, the future path table of each AGV is screened to obtain the combined passage mode of AGVs without conflicts, including:

[0119] (1) For each intersection, determine whether there is a conflict in the passage of the AGV based on the future path table of each AGV;

[0120] (2) If it is determined that there is an AGV combination with a passage conflict, the AGV combination with the passage conflict is removed, and a passage mode is obtained without the AGV combination with the passage conflict.

[0121] In an embodiment of the present invention, judging whether there is a conflict in the passage of the AGV according to the future path table of each AGV includes:

[0122] For two AGVs, determine the future path sets corresponding to each AGV as R1 and R2 respectively;

[0123] The minimum distance between the future paths of two AGVs is calculated according to the minimum distance calculation formula, where the minimum distance calculation formula is:

[0124] ,

[0125] in, represents the minimum distance between all elements in R1 and all elements in R2, Indicates the minimum safe distance between two AGVs;

[0126] If the minimum distance formula is satisfied, it is determined that there is no conflict between the two AGVs;

[0127] If the minimum distance formula is not satisfied, it is determined that there is a conflict between the two AGVs.

[0128] It should be understood that for Two AGVs in and , and its corresponding future path set is and , judged by the following formula and Is there a conflict:

[0129] ,

[0130] in, express All elements in and All elements in The minimum distance between Indicates the minimum safe distance between AGVs. If the above conflict judgment formula is satisfied, it means that AGV and There is no conflict between them, that is, they can be assigned the same traffic priority and pass through the intersection at the same time.

[0131] S230, finding the best passage sequence for each intersection in the AGV combined passage mode according to the MCTS algorithm and the optimization target of the intersection passage model;

[0132] In the embodiment of the present invention, the search for the best passing order is achieved by constructing a search tree based on the MCTS algorithm and performing iterations.

[0133] Specifically, according to the MCTS algorithm and the optimization target of the intersection traffic model, the optimal traffic sequence of each intersection in the AGV combined traffic mode is found, such as Figure 5 As shown, including:

[0134] S231, constructing a search tree for each intersection, wherein the search tree includes a root node and multiple layers of nodes sequentially extending from the root node, each node includes a combination of AGVs that can pass through at the same time, and each node can expand downward to a child node, the node at the first layer has the highest priority, and each branch of the search tree starting from the root node represents a feasible passage order;

[0135] In the embodiment of the present invention, the intersection Construct a search tree. Each node on the tree includes the AGV combination that can pass through at the same time. Each node will expand downward to a child node. The nodes at the first level have the highest priority. Each branch of the search tree starting from the root node represents a feasible passage order.

[0136] S232, iterating the search tree according to the MCTS algorithm until the optimization target of the intersection traffic model is met, and obtaining the optimal traffic sequence for each intersection in the AGV combined traffic mode.

[0137] In the embodiment of the present invention, MCTS is applied to find the best pass order. MCTS iteratively constructs the search tree, and one iteration consists of four steps: selection, expansion, simulation, and back propagation. Figure 6 As shown in the figure, it is a schematic diagram of the basic process of constructing a search tree for MCTS.

[0138] Specifically, iterating the search tree according to the MCTS algorithm includes:

[0139] a1. Starting from the root node, select the node to be expanded by using the simulation strategy formula, wherein the simulation strategy formula is expressed as:

[0140] ,

[0141] in, Represents a child node The score, Indicates the number of times the current node has been visited. Represents a child node The number of times it has been visited. represents the weight coefficient;

[0142] In the embodiment of the present invention, starting from the root node, the node to be expanded is selected through the simulation strategy of the above simulation strategy formula, where the first term encourages the use of known nodes and the second term encourages the exploration of unvisited nodes.

[0143] a2. According to the nodes selected by the simulation strategy formula, new nodes are gradually expanded downward to reach the target state, where the target state is a state that can include all elements of the AGV set;

[0144] It should be understood that based on the nodes selected in the previous step, new nodes are gradually expanded downward to reach the target state, which needs to include All elements in the collection, that is, building a complete branch.

[0145] a3. Calculate the score of the current node;

[0146] In the embodiment of the present invention, calculating the score of the current node includes:

[0147] 1) Determine the first part of the score according to the branch level where the current node is located. The calculation formula for the first part of the score is:

[0148] ,

[0149] in, represents the first part of the score, Indicates the level of the branch where the current node is located. represents the corresponding weight coefficient;

[0150] It should be understood that in order to improve the traffic efficiency of multiple AGVs, the present invention defines the AGV set included in the current node as , The number of elements in is For intersections ,At the same time, the fewer the branch layers the current node is located at, the more AGVs can pass through at the same time.

[0151] 2) The second part of the score is determined according to the number of AGVs included in the current node and the number of AGVs queued behind it. The calculation formula for the second part of the score is:

[0152] ,

[0153] in, represents the second part of the score, Indicates the AGV set included in the current node Middle The number of AGVs that queue up after each element, represents the corresponding weight coefficient;

[0154] In the embodiment of the present invention, the second part score is calculated by the number of AGVs queued after the AGV included in the current node, and the queued AGVs are obtained and the AGVs that are too far from the intersection area are screened out by setting the queue distance threshold. The more AGVs are queued later, the higher the score of this part of the current node.

[0155] 3) The third part score is determined according to the coordinated impact results of adjacent intersections. The calculation formula of the third part score is:

[0156] ,

[0157] in, represents the third part of the score, Indicates the current intersection. express The adjacent intersections, Indicates that you are currently in The next intersection is And through The lane and The same total number of AGVs, Indicates the AGV set included in the current node Middle AGVs, Indicates the distance between the adjacent intersection and the next intersection. represents the corresponding weight coefficient;

[0158] In the embodiment of the present invention, in order to avoid congestion, the influence of the result of the coordination method between adjacent intersections on the coordination processing of adjacent intersections will be considered. The collection Elements , if the intersection Among the adjacent intersections of AGV, and it will pass The lane and If the current lane is the same, then The score of this part of the node will be calculated according to the calculation formula in the third part above.

[0159] In the calculation formula of the third part, Indicates the current intersection. express The adjacent intersections, Indicates that you are currently in The next intersection is And through The lane and The same total number of AGVs, Indicates the AGV set included in the current node Middle AGVs; express and If and The distance between them is large. The AGV drives to When nearby, May have left .

[0160] 4) The score of the current node is calculated based on the first, second and third part scores. The calculation formula is:

[0161] .

[0162] a4. Back-propagate based on the score of the current node to update the score of the parent node.

[0163] It should be understood that the simulation results will be back-propagated from the current node, and the scores of the parent nodes will be updated accordingly.

[0164] S240: Determine the passage priority of each AGV at each intersection according to the optimal passage sequence.

[0165] In the embodiment of the present invention, the traffic priority of each AGV is sent to the corresponding AGV, and each AGV can perform the transportation operation according to its own traffic priority. Specifically, the AGV is assigned a task and starts to drive and perform the transportation operation; at regular intervals, the traffic priority is assigned to the AGV waiting to pass through the control area of ​​each intersection. The steps of assigning traffic priority are repeated until all AGVs complete the operation and stop.

[0166] The following is combined with Figure 7 To Attachment Fig. 9 The specific process of the multi-AGV intersection coordination method for an automated container terminal provided by the present invention is described in detail.

[0167] like Figure 7 As shown in the figure, this is an example of considering the first part of the score. In the left figure, three AGVs A, B, and C are waiting to pass through the intersection. Their trajectories inside the intersection are shown by arrows. There are trajectory conflicts between A and C, and between B and C. Therefore, in the search tree shown in the right figure, there is no node containing both AC and BC. Since there is no conflict between the trajectories of A and B, the two AGVs can pass through the intersection at the same time. Therefore, the first part of the score corresponding to the node containing AB in the search tree will be greater than that of other nodes, and the priority will therefore be higher.

[0168] like Figure 8 As shown in FIG. 1 , an example of considering the second part of the score is shown. Three AGVs A, B, and C are waiting to pass through the intersection, and AGV D is queued behind B, and D meets the queue distance threshold. Therefore, the second part score of the node containing B in the search tree will be greater than that of other nodes, and the priority will be increased accordingly.

[0169] like Fig. 9 As shown, this is an example considering the scoring situation in the third part. There are three AGVs A, B, and C waiting to pass at the intersection, and A comes from an adjacent intersection. , B comes from the adjacent intersection , C comes from the adjacent intersection . At this time, D passes by, and its next intersection is ,pass The lane of is the same as A. Similarly, At this time, E passes, and its next intersection is ,pass The lane of A is the same as B; At this time, F passes by, and its next intersection is ,pass The lane of is the same as that of C. represents A, Indicates B, Represents C. According to the scoring formula in the third part, at this time, are equal to 1. However, due to and The distance between them is far, so although the nodes including A, B, and C all have the third part score, the score of the node including B will be lower than the score of the nodes including A and B, and the priority will be lower. After comprehensively considering the three parts of the score, the embodiment of the present invention determines the best passage order based on the comprehensive score of the three parts.

[0170] In summary, the automated container terminal multi-AGV intersection coordination method provided by the present invention, in the automated container terminal scenario, models the intersection as a polygonal area, takes into account the geometric properties of the intersection and the driving route of the AGV in the intersection, and allows more than one AGV to pass through the intersection at the same time. The control area is set by extending the intersection area outward by a certain distance, and the AGVs that arrive at the control area are added to the set of AGVs waiting to pass, and the AGVs that leave the intersection area are deleted from the set of AGVs waiting to pass.

[0171] In addition, the automated container terminal multi-AGV intersection coordination method of the present invention is periodically applied to the intersection, and a traffic priority is assigned to all elements in the AGV set waiting to pass through the intersection, and the AGV passes through the intersection according to the traffic priority. The conflicting AGV combination is filtered out through the future path of the AGV to reduce search redundancy. The traffic priority is periodically assigned to the AGVs waiting to pass through the intersection. After considering the three aspects of simultaneous passage of multiple AGVs, queue relief and congestion avoidance, the MCTS is comprehensively improved and applied to search for the best traffic order, and conflict-free deadlock and efficient multi-AGV intersection coordinated passage is achieved.

[0172] It is to be understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of the present invention, but the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A multi-AGV intersection coordination method for an automated container terminal, characterized in that: include: Determine the current number of intersections and the number of AGVs, and construct an intersection traffic model according to the current number of intersections and the number of AGVs, wherein the intersection traffic model at least includes a polygonal intersection model; According to the MCTS algorithm, the traffic order of the intersection traffic model is searched to obtain the traffic priority of each AGV; The traffic priority of each AGV is sent to the corresponding AGV, and each AGV can perform transportation operations according to its own traffic priority; The traffic order of the intersection traffic model is searched according to the MCTS algorithm to obtain the traffic priority of each AGV, including: According to the MCTS algorithm, the optimization goal of the intersection traffic model is to minimize the average AGV intersection passing time; For each intersection, the future path table of each AGV is filtered to obtain the combined passage mode of AGVs without conflicts. Find the best traffic sequence for each intersection in the AGV combined traffic mode according to the MCTS algorithm and the optimization target of the traffic model at the intersection; Determine the passage priority of each AGV at each intersection according to the optimal passage order; According to the MCTS algorithm and the optimization target of the intersection traffic model, the optimal traffic sequence of each intersection in the AGV combined traffic mode is found, including: A search tree is constructed for each intersection, wherein the search tree includes a root node and multiple layers of nodes extending sequentially from the root node, each node includes a combination of AGVs that can pass through at the same time, and each node can expand downward to a child node, the node at the first layer has the highest priority, and each branch of the search tree starting from the root node represents a feasible passage order; Iterate the search tree according to the MCTS algorithm until the optimization goal of the intersection traffic model is met, and obtain the optimal traffic sequence of each intersection in the AGV combined traffic mode; Iterating the search tree according to the MCTS algorithm includes: Starting from the root node, a node to be expanded is selected by a simulation strategy formula, wherein the simulation strategy formula is expressed as: ， in, Represents a child node The score, Indicates the number of times the current node has been visited. Represents a child node The number of times it has been visited. represents the weight coefficient; According to the nodes selected by the simulation strategy formula, new nodes are gradually expanded downward to reach the target state, wherein the target state is a state that can include all elements of the AGV set; Calculate the score of the current node; Back-propagate based on the score of the current node to update the score of the parent node; For each intersection, the future path table of each AGV is filtered to obtain the combined passage mode of AGVs without conflicts, including: For each intersection, determine whether there is a conflict in the passage of AGVs based on the future path table of each AGV; If it is determined that there is an AGV combination with a passage conflict, the AGV combination with the passage conflict is removed to obtain a passage mode without the AGV combination with the passage conflict; Determine whether there is a conflict in the passage of the AGV based on the future path table of each AGV, including: For two AGVs, determine the future path sets corresponding to each AGV as R1 and R2 respectively; The minimum distance between the future paths of two AGVs is calculated according to the minimum distance calculation formula, where the minimum distance calculation formula is: ， in, represents the minimum distance between all elements in R1 and all elements in R2, Indicates the minimum safe distance between two AGVs; If the minimum distance calculation formula is satisfied, it is determined that there is no conflict between the two AGVs; If the minimum distance calculation formula is not satisfied, it is determined that there is a conflict between the two AGVs; Calculate the score of the current node, including: The first part of the score is determined according to the branch layer number where the current node is located. The calculation formula of the first part of the score is: ， in, represents the first part of the score, Indicates the level of the branch where the current node is located. represents the corresponding weight coefficient; The second part of the score is determined according to the number of AGVs included in the current node and the number of AGVs queued behind it. The calculation formula of the second part of the score is: ， in, represents the second part of the score, Indicates the AGV set included in the current node Middle The number of AGVs that queue up after each element, represents the corresponding weight coefficient, Represents an AGV collection The number of elements in ; The third part score is determined according to the collaborative impact results of adjacent intersections. The calculation formula of the third part score is: ， in, represents the third part of the score, Indicates the current intersection. express The adjacent intersections, Indicates that you are currently in The next intersection is And through The lane and The same total number of AGVs, Indicates the AGV set included in the current node Middle AGVs, Indicates the distance between the adjacent intersection and the next intersection. represents the corresponding weight coefficient; The score of the current node is calculated based on the first part score, the second part score and the third part score. The calculation formula is: 。 2. The method for coordinating multiple AGV intersections at an automated container terminal according to claim 1, characterized in that: Determine the current number of intersections and AGVs, and build an intersection traffic model based on the current number of intersections and AGVs, including: Get all current intersection information and all current AGV information; Generate an intersection list according to all current intersection information, and generate an AGV list and an AGV future path list according to all current AGV information, wherein each AGV future path list includes multiple waypoint information; Determine the set of AGVs that are waiting to pass through at any intersection and at the same time for each AGV based on the AGV future path list; A parking area model for an intersection is constructed, and a traffic priority set for the AGV set is constructed.

3. The method for coordinating multiple AGV intersections at an automated container terminal according to claim 2, characterized in that: Construct a parking area model at an intersection, including: Construct a polygonal model matching the shape of the intersection according to the shape of the intersection; A peripheral control line is arranged at a preset distance from the polygonal area on the periphery of the polygonal model, and a control area is formed between the peripheral control line and the polygonal model; When each AGV enters the control area of ​​the current intersection, the AGV is added to the AGV set, and when the AGV leaves the control area of ​​the current intersection, the AGV is deleted from the AGV set.

4. The method for coordinating multiple AGV intersections at an automated container terminal according to claim 2, characterized in that: Constructing a traffic priority set of the AGV set, including: Determine the real-time assigned traffic priority of each AGV in the AGV set; A traffic priority set of the AGV set is determined according to the real-time assigned traffic priority of each AGV.

Images