A Shuttle Traffic Control Method and System Based on Improved A* Algorithm Path Planning

Abstract

This invention discloses a shuttle traffic control method and system based on an improved A* algorithm path planning. The method involves acquiring and parsing a task to obtain its start and end points; using the improved A* algorithm to plan the optimal path from the start to the end point; obtaining path occupancy records from the scheduling system and implementing segmented occupancy management for the optimal path. Occupancy management includes currently exclusive mobile occupancy and reserved occupancy marked for pre-reservation. The shuttle moves along the optimal path, and upon reaching a reserved path segment, it moves to the optimal waiting point and triggers a passage request. During the shuttle's movement, previously traversed path resources are released in real time. The improved A* algorithm enhances path search efficiency, and the segmented path occupancy management traffic control strategy significantly reduces path conflicts.

Description

Technical Field

[0001] This invention relates to intelligent warehousing, specifically to a shuttle traffic control method and system based on improved A* algorithm path planning. Background Technology

[0002] With the increasing automation of logistics, the workload in automated storage and retrieval systems (AS / RS) is increasing dramatically, as is the number of four-way shuttles on each level. This leads to a surge in demands for the performance and efficiency of four-way shuttle path planning and traffic control. Current technical solutions primarily employ static path calculation and whole-segment path locking strategies for four-way shuttle path planning, which are inefficient in scenarios involving multiple four-way shuttles operating collaboratively.

[0003] For example, in patent application number 202511391536.5, a predetermined path is reserved. When path intersections or conflicts occur, a first-come, first-served approach is used to lock the path. Severe path conflicts can easily lead to deadlocks. Another example is patent application number 202511804415.9, which uses the A* algorithm based on static paths. However, when conflicts occur, recalculation is required. Because the A* algorithm has a large computational load and poor real-time performance, it is difficult to meet the needs of large-scale warehousing systems, resulting in low planning efficiency. Yet another example is patent application number 202510663221.5, which calculates paths by combining equipment operating parameters and virtual sensors. However, considering the performance of different equipment, warehouse environment, and load conditions, it cannot adapt to various scenarios and has poor scalability. Summary of the Invention

[0004] Objective: To address the above-mentioned shortcomings, this invention provides a shuttle traffic control method based on an improved A* algorithm path planning, which enhances planning efficiency and scalability. This invention also provides a shuttle traffic control system based on an improved A* algorithm path planning.

[0005] Technical Solution: To solve the above problems, the present invention provides a shuttle traffic control method based on improved A* algorithm path planning, comprising the following steps:

[0006] When the shuttle is idle, it actively requests a task and parses the requested task to obtain the start and end points of the task. An improved A* algorithm is used to plan the optimal path from the start to the end point. The heuristic function of the improved A* algorithm comprehensively considers the accessibility of neighboring points and the Manhattan distance between the current node and the end point.

[0007] The path occupancy record is obtained from the scheduling system to obtain the path occupancy status. Based on the path occupancy status, the optimal path is segmented and managed in the scheduling system. The occupancy management includes moving occupancy and reserved occupancy. Moving occupancy means that the current device has exclusive access to the path and there is no conflict. Reserved occupancy means that when the path is occupied by other devices, the current device reserves the path and finds the optimal waiting point.

[0008] The shuttle receives the optimal path from the dispatch system, which includes information on moving occupancy and reserved occupancy. The shuttle moves along the optimal path. When it reaches the reserved occupancy section of the path, it moves to the optimal waiting point and triggers a passage request. The dispatch system then determines whether to allow passage.

[0009] The shuttle communicates with the scheduling system in real time during its movement and releases the path occupancy records of the traversed paths in the scheduling system in real time.

[0010] Furthermore, the improved A* algorithm calculates static paths based on the device's tolerance for path conflicts. and dynamic paths The sum of the actual costs between all nodes on a dynamic path is used as the path weight, while the sum of the actual costs between all nodes on a static path, combined with the task waiting time, is used as the path weight. The optimal path is determined based on minimizing the path weight. The expression is:

[0011]

[0012] in, W represents the waiting time factor, and W is the task waiting time weight.

[0013] The static path To accept the results of path planning in the event of path conflicts, the The result of route planning to avoid path conflicts.

[0014] Furthermore, the cost function of the improved A* algorithm is:

[0015]

[0016]

[0017]

[0018] in, This is the actual cost function; For heuristic cost function; The comprehensive cost function; The path occupancy penalty coefficient is used to represent the device's tolerance for path conflicts. The value range is [0, 2]. When it is 0, it means that the occupancy of other devices is not considered, which is a static weight; when it is 2, it means that the device is extremely averse to path conflicts with other devices, which is a dynamic weight. Path occupancy penalty; This is the steering penalty coefficient; For turning penalty; This refers to the actual path weight coefficient; This represents the actual path weight; Manhattan distance coefficient; This represents the Manhattan distance from the current node to the destination.

[0019] Furthermore, when the scheduling system receives a passage request, it determines whether the movement occupancy record of the requested path has been released. If the movement occupancy record has been released, the scheduling system replies with a permission signal and marks the path as movement occupancy again.

[0020] Furthermore, the method for determining the optimal waiting point is as follows: based on the optimal path, reverse the search for the non-conflict position before the current conflict path. The non-conflict position includes a node at a preset distance from the starting conflict point, or a node representing the intersection position.

[0021] Furthermore, when the waiting time of the shuttle at the optimal waiting point exceeds a preset threshold, the shuttle will be re-planned, and the static and dynamic paths will be recalculated to avoid deadlock; among them, the weight of the static path will increase dynamically as the equipment waiting time increases.

[0022] Furthermore, the scheduling system acquires the location and task status of all shuttles in real time, calculates the optimal path, and then confirms whether there are other idle shuttles on the optimal path. If so, the scheduling system directly issues a movement command to the idle shuttle to control it to avoid the obstacle.

[0023] Furthermore, when parsing the requested job tasks, a topological map model of the warehouse is established, and the warehouse layout is abstracted into a point topological graph structure. The topological map model includes nodes and edges. Nodes represent location points, and edges represent feasible paths. Path weights are constructed based on real-time path occupancy, turning costs, and cargo status.

[0024] The present invention discloses a shuttle traffic control system based on an improved A* algorithm path planning, comprising:

[0025] The task request module is used to proactively request work tasks when the shuttle is idle;

[0026] The task parsing module is used to parse the requested job tasks and obtain the start and end points of the job tasks;

[0027] The path planning module is used to plan the optimal path from the starting point to the ending point using an improved A* algorithm; the heuristic function of the improved A* algorithm comprehensively considers the descentability of neighboring points and the Manhattan distance between the current node and the ending point.

[0028] The conflict detection module is used to obtain path occupancy records from the scheduling system to determine the path occupancy status. Based on the path occupancy status, it performs segmented occupancy management of the optimal path in the scheduling system. The occupancy management includes moving occupancy and reserved occupancy. Moving occupancy refers to the current device exclusively occupying the path without conflict. Reserved occupancy refers to the current device marking the path as reserved when it is occupied by other devices and finding the optimal waiting point. During the shuttle's movement, it communicates with the scheduling system in real time and releases the path occupancy records of the traversed paths in the scheduling system in real time.

[0029] The scheduling execution module is used to receive the optimal path containing moving and reserved occupancy information issued by the scheduling system, control the shuttle to move along the optimal path, and when it reaches the reserved occupancy path segment, control the shuttle to move to the optimal waiting point and trigger the passage request.

[0030] Furthermore, it also includes a map modeling module, which is used to build a topological map model of the warehouse, abstracting the warehouse layout into a point topological graph structure. The topological map model includes nodes and edges, where nodes represent location points and edges represent feasible paths. Path weights are constructed based on real-time path occupancy, turning costs, and cargo status.

[0031] Beneficial Effects: Compared to existing technologies, the significant advantages of this invention are: improved path search efficiency by over 30% through the improved A* algorithm; and significantly reduced path conflicts through a segmented path occupancy management traffic control strategy. Furthermore, based on dynamic path weights, the path conflict rate is reduced to below 40%, a significant reduction. The optimized scheduling strategy increases system throughput by 25%, intelligent path planning reduces average energy consumption by 15%, equipment utilization increases from 60% to 85%, reduces on-site operators by 50%, saving labor costs, and increases order processing speed by 40%, thus improving operational efficiency. The improved A* algorithm is independent of external factors such as equipment performance and warehouse environment, and can be ported to different automated warehouse systems, exhibiting good scalability. Attached Figure Description

[0032] Figure 1 This is a flowchart illustrating the planning method in this invention. Detailed Implementation

[0033] Example 1

[0034] like Figure 1 As shown in the figure, a shuttle traffic control method based on improved A* algorithm path planning in this embodiment includes the following steps:

[0035] When the shuttle is idle, it actively requests work tasks and parses the requested tasks to obtain the start and end points. During task parsing, map modeling is performed: the warehouse layout is abstracted into a point topology graph structure, where nodes represent locations and edges represent feasible paths. Path weights are constructed based on real-time path occupancy, turning costs, and cargo status.

[0036] An improved A* algorithm is used to plan the optimal path from the starting point to the ending point; the heuristic function of the improved A* algorithm comprehensively considers the descentability of neighboring points and the Manhattan distance between the current node and the ending point.

[0037] The path occupancy record is obtained from the scheduling system to obtain the path occupancy status. Based on the path occupancy status, the optimal path is segmented and managed in the scheduling system. The occupancy management includes moving occupancy and reserved occupancy. Moving occupancy means that the current device has exclusive access to the path and there is no conflict. Reserved occupancy means that when the path is occupied by other devices, the current device reserves the path and finds the optimal waiting point.

[0038] The shuttle receives the optimal path from the dispatch system, which includes information on moving occupancy and reserved occupancy. The shuttle moves along the optimal path. When it reaches the reserved occupancy section of the path, it moves to the optimal waiting point and triggers a passage request. The dispatch system then determines whether to allow passage.

[0039] The shuttle communicates with the scheduling system in real time during its movement and releases the path occupancy records of the traversed paths in the scheduling system in real time.

[0040] In this embodiment, traffic control is implemented on the equipment based on path conflict, including strategies such as waiting and avoidance; for paths with long waiting times, the routes are replanned to avoid conflict deadlock.

[0041] This embodiment provides an improved path planning logic for the A* algorithm. The algorithm defines the state of a node as a combination of "position + direction" and calculates the globally optimal path, including turning energy consumption, by introducing a turning penalty mechanism.

[0042] 1. Definition of the core data structure of the algorithm

[0043] This algorithm defines two core internal data structures: state and search node.

[0044] Status: Used to uniquely identify a point in the graph.

[0045]

[0046] in, For node indexing, This indicates the direction when entering this node.

[0047] Search node: An element used for sorting in the priority queue, containing its state. Actual cost Heuristic Costs and comprehensive costs .

[0048] 2. Path planning execution steps

[0049] The specific execution flow of the algorithm is as follows:

[0050] Step S1: Initialize parameters.

[0051] Input: Starting point ,end The initial direction of the current vehicle Initialize the priority queue and the set of visited states; initialize the cost map, which records the minimum cost to reach each state; initialize the predecessor map, which records path backtracking.

[0052] Create starting state And add it to the priority queue, at this time .

[0053] Step S2: Loop search (main loop).

[0054] When the priority queue is not empty, perform the following operations:

[0055] S2.1 Pop Node: Pop a node from the priority queue The node with the smallest value is denoted as the current node.

[0056] S2.2 Target Detection: If the index of the current node is equal to the target index (i.e., the endpoint) If the path is not found, proceed to step S3 (path backtracking).

[0057] S2.3 Neighbor Expansion: Traverse all neighboring nodes v of the current node x. The original A* algorithm does not record the walkable neighbor nodes of a point, so it must traverse all nodes, and its algorithm complexity is O(n log n). The complexity of this algorithm is reduced to... .

[0058] If the weight of edge (x,v) is invalid, skip it.

[0059] Get the direction of movement from x to v .

[0060] S2.4 Calculate path cost:

[0061]

[0062]

[0063]

[0064] in, For the actual cost, For the cost of heuristics, For the overall cost, This is the path occupancy penalty coefficient, used to control the impact of path occupancy on path calculation, representing the device's tolerance for path conflicts. The value ranges from [0,2]. A value of 0 indicates that occupancy by other devices is not considered, i.e., static weight; a value of 2 indicates an extreme aversion to path conflicts with other devices. The path occupancy penalty is calculated based on the occupancy type, the current device's path direction, and the direction of paths occupied by other devices. This is the steering penalty coefficient, used to control the impact of steering on the path calculation. Its value ranges from [0.5, 1] ​​and should be set according to the actual load and warehouse environment. The turning penalty is calculated based on whether the current node's "entry direction" and "exit direction" are perpendicular. If they are perpendicular, the value is 5; otherwise, the value is 0. This is the actual path weight coefficient, which is usually set to 1. This represents the actual path weight. is the Manhattan distance coefficient, used to control the influence of the heuristic function on the path calculation, with a value range of [0.8, 1]. This represents the Manhattan distance from the current node to the target node.

[0065] S2.5 Select the most suitable path:

[0066] The path weight is calculated by summing the actual costs between each node on the path, based on the static path calculated in step S2.4. and dynamic paths The total actual cost between each node on the dynamic path is used as the path weight of the dynamic path, and the total actual cost between each node on the static path is combined with the task waiting time as the path weight of the static path. The path with the smallest path weight is then determined.

[0067]

[0068]

[0069]

[0070] in, For the first dynamic path A real cost, For the first static path A real cost, The waiting time factor determines the impact of waiting time on the optimal path, with a value range of [0, 1]; W is the task waiting time weight. , The average speed of the equipment. The time when the task was issued. This is the current time.

[0071] Dynamic paths consider path conflicts when calculating path costs. Therefore, the calculated path may bypass other conflicting devices and take a longer route, but without waiting. The final path weight of a dynamic path is represented by the sum of the actual costs between all nodes on the path. Static paths, on the other hand, are the shortest paths from the starting point to the destination. Path conflicts are not considered when calculating path costs. This means that there may be other devices on the planned path, and you need to wait for them to leave the path before you can proceed. The final path weight of a static path is represented by the sum of the actual costs between all nodes on the path combined with the waiting time of the shuttle.

[0072] Step S3: Path backtracking.

[0073] Starting from the endpoint state, trace back to the starting state level by level through the predecessor mapping table. Convert the traced state sequence into an actual list of nodes.

[0074] Step S4: Path smoothing and merging (post-processing).

[0075] Traverse the actual list of nodes, merge paths based on the direction and occupancy of the paths formed by the preceding and following nodes, and generate the final driving instruction.

[0076] In this embodiment, the improved A* algorithm performs real-time path weight optimization, constructing actual path weights based on the loading status of equipment and storage locations within the same area; it optimizes the heuristic function by considering the Manhattan distance between the current point and the target point, improving performance by over 30%; and it optimizes the loop pruning by considering the traversability of the current point's neighboring points and optimizing the node traversal logic, reducing the algorithm complexity from... Reduce to Perform multi-objective optimization: simultaneously consider multiple optimization objectives such as path length, time cost, and turning energy consumption.

[0077] This embodiment employs a multi-level traffic control strategy:

[0078] Conflict avoidance: When calculating the path, consider the real-time location and path occupancy of other devices, and try to avoid conflicting paths.

[0079] Idle device avoidance: When the optimal path is obtained, if there are idle devices on the path, the idle device avoidance task is directly issued.

[0080] Optimal waiting point: When the optimal path is obtained, if other devices occupy the path, the optimal waiting point ahead of the first conflicting path is searched in reverse. The optimal waiting point can be configured, such as the first 3 points or the previous intersection.

[0081] Passage Inquiry: After the device moves to the waiting point, it inquires with the scheduling system whether it can pass. The scheduling system determines whether there is a conflict on the path of the current single movement command. If not, it allows passage.

[0082] This embodiment is based on a planning strategy with controllable waiting time:

[0083] Dual path planning: Calculates the optimal path based on both dynamic and static path weights, increasing the static path weight based on device latency. This avoids both temporary occupancy by other devices causing the current device to take a longer route and prolonged latency caused by other devices.

[0084] Timeout optimization: For paths that have been waiting for a long time, replan them to avoid conflicts and deadlocks.

[0085] This embodiment uses a heavy-duty four-way vehicle warehouse as an example. The warehouse has 5 floors, with 250 storage locations on each floor. The standard storage location width is 1.2m. The four-way vehicles have a load capacity of 1 ton, an average moving speed of 1.2m / s, and a reversing time of 3s.

[0086] The specific implementation steps are as follows:

[0087] When the four-way shuttle is idle, it actively requests work tasks, selects executable tasks and assigns them to idle four-way shuttles. The four-way shuttle then parses the work tasks. The warehouse layout is abstracted into a point topology graph structure, and a topology map model of the warehouse is constructed. The topology map model includes nodes and edges. Nodes represent location points, and location points can be set for storage locations, main walking paths, and sub-walking paths. Edges represent feasible paths. Based on real-time path occupancy, turning costs, and cargo status, path weights are constructed. The start and end points of the work tasks in the topology map model are obtained.

[0088] An improved A* algorithm is used to find the optimal path. The algorithm parameters are specifically set as follows: path occupancy penalty coefficient. Static weights are set to 0, while dynamic weights are set to 1; steering penalty coefficient. Set to 0.8; actual path weight coefficient Set to 1; Manhattan distance coefficient Set to 1; Waiting time factor Set it to 0.9.

[0089] After obtaining a path, the entire path is reserved. There are two types of reservation: moving reservation, where the current device has exclusive access and there is no path conflict; and reserved reservation, where other devices have already reserved the path, and the current device marks the path as reserved and finds the optimal waiting point. Unit paths with the same reservation type and the same direction are merged into a complete path, ultimately forming a set of device instructions.

[0090] The planned route instructions are issued to the four-way shuttle. During execution, the shuttle releases the routes it has already traversed in real time. When traversing a reserved route, it moves to the optimal waiting point and triggers a passage request, which requires querying the scheduling system to confirm passage. For routes with long waiting times, the route is replanned to avoid conflicts and deadlocks.

[0091] Example 2

[0092] This embodiment presents a shuttle traffic control system based on an improved A* algorithm path planning, comprising:

[0093] The task request module is used to proactively request work tasks when the shuttle is idle;

[0094] Task assignment module: Used to assign tasks to appropriate four-way shuttles;

[0095] The task parsing module is used to parse the requested job tasks and obtain the start and end points of the job tasks;

[0096] Map modeling module: Used to create a topological map model of the warehouse;

[0097] Path planning module: used to calculate the optimal path;

[0098] Conflict detection module: Used to detect and resolve path conflicts, and to reserve and release paths.

[0099] Scheduling and execution module: Used to control the movement and command execution of the four-way shuttle.

Claims

1. A shuttle traffic control method based on improved A* algorithm path planning, characterized in that, Includes the following steps: When the shuttle is idle, it actively requests a task and analyzes the requested task to obtain the start and end points of the task. An improved A* algorithm is used to plan the optimal path from the starting point to the ending point; the heuristic function of the improved A* algorithm comprehensively considers the descentability of neighboring points and the Manhattan distance between the current node and the ending point. The path occupancy record is obtained from the scheduling system to obtain the path occupancy status. Based on the path occupancy status, the optimal path is segmented and managed in the scheduling system. The occupancy management includes moving occupancy and reserved occupancy. Moving occupancy means that the current device has exclusive access to the path and there is no conflict. Reserved occupancy means that when the path is occupied by other devices, the current device reserves the path and finds the optimal waiting point. The shuttle receives the optimal path from the dispatch system, which includes information on moving occupancy and reserved occupancy. The shuttle moves along the optimal path. When it reaches the reserved occupancy section of the path, it moves to the optimal waiting point and triggers a passage request. The dispatch system then determines whether to allow passage. The shuttle communicates with the scheduling system in real time during its movement and releases the path occupancy records of the traversed paths in the scheduling system in real time.

2. The shuttle traffic control method based on improved A* algorithm path planning according to claim 1, characterized in that, The improved A* algorithm calculates static paths based on the device's tolerance for path conflicts. and dynamic paths The sum of the actual costs between all nodes on a dynamic path is used as the path weight, while the sum of the actual costs between all nodes on a static path, combined with the task waiting time, is used as the path weight. The optimal path is determined based on minimizing the path weight. The expression is: in, W represents the waiting time factor, and W is the task waiting time weight. The static path To accept the results of path planning in the event of path conflicts, the The result of route planning to avoid path conflicts.

3. The shuttle traffic control method based on improved A* algorithm path planning according to claim 2, characterized in that, The cost function of the improved A* algorithm is: in, This is the actual cost function; For heuristic cost function; The comprehensive cost function; The path occupancy penalty coefficient is used to represent the device's tolerance for path conflicts. The value range is [0, 2]. When it is 0, it means that the occupancy of other devices is not considered, which is a static weight; when it is 2, it means that the device is extremely averse to path conflicts with other devices, which is a dynamic weight. Path occupancy penalty; This is the steering penalty coefficient; For turning penalty; This refers to the actual path weight coefficient; This represents the actual path weight; Manhattan distance coefficient; This represents the Manhattan distance from the current node to the destination.

4. The shuttle traffic control method based on improved A* algorithm path planning according to claim 3, characterized in that, When the scheduling system receives a passage request, it determines whether the movement occupancy record of the requested path has been released. If the movement occupancy record has been released, the scheduling system replies with a permission signal and marks the path as movement occupancy again.

5. The shuttle traffic control method based on improved A* algorithm path planning according to claim 4, characterized in that, The method for determining the optimal waiting point is as follows: based on the optimal path, reverse the search for the non-conflict position before the current conflict path. The non-conflict position includes a node at a preset distance from the starting conflict point, or a node representing the intersection position.

6. The shuttle traffic control method based on improved A* algorithm path planning according to claim 5, characterized in that, When the waiting time of the shuttle at the optimal waiting point exceeds a preset threshold, the shuttle will be re-planned, and the static and dynamic paths will be recalculated to avoid deadlock. The weight of the static path will increase dynamically as the waiting time increases.

7. The shuttle traffic control method based on improved A* algorithm path planning according to claim 6, characterized in that, The scheduling system obtains the location and task status of all shuttles in real time, calculates the optimal path, and then checks whether there are other idle shuttles on the optimal path. If so, the scheduling system directly issues a movement command to the idle shuttle to control it to avoid the obstacle.

8. The shuttle traffic control method based on improved A* algorithm path planning according to claim 1, characterized in that, When parsing the requested job tasks, a topological map model of the warehouse is established, and the warehouse layout is abstracted into a point topological graph structure. The topological map model includes nodes and edges. Nodes represent location points, and edges represent feasible paths. Path weights are constructed based on real-time path occupancy, turning costs, and cargo status.

9. A shuttle traffic control system based on improved A* algorithm path planning, characterized in that, include: The task request module is used to proactively request work tasks when the shuttle is idle; The task parsing module is used to parse the requested job tasks and obtain the start and end points of the job tasks; The path planning module is used to plan the optimal path from the starting point to the destination using an improved A* algorithm; the heuristic function of the improved A* algorithm comprehensively considers the accessibility of neighboring points and the Manhattan distance between the current node and the destination. The conflict detection module is used to obtain path occupancy records from the scheduling system to determine the path occupancy status. Based on the path occupancy status, it performs segmented occupancy management of the optimal path in the scheduling system. The occupancy management includes moving occupancy and reserved occupancy. Moving occupancy refers to the current device exclusively occupying the path without conflict. Reserved occupancy refers to the current device marking the path as reserved when it is occupied by other devices and finding the optimal waiting point. During the shuttle's movement, it communicates with the scheduling system in real time and releases the path occupancy records of the traversed paths in the scheduling system in real time. The scheduling execution module is used to receive the optimal path containing moving and reserved occupancy information issued by the scheduling system, control the shuttle to move along the optimal path, and when it reaches the reserved occupancy path segment, control the shuttle to move to the optimal waiting point and trigger the passage request.

10. The shuttle traffic control system based on improved A* algorithm path planning according to claim 9, characterized in that, It also includes a map modeling module, which is used to build a topological map model of the warehouse, abstracting the warehouse layout into a point topological graph structure. The topological map model includes nodes and edges, where nodes represent location points and edges represent feasible paths. Path weights are constructed based on real-time path occupancy, turning costs, and cargo status.

Images