A multi-agv continuous task path planning method and system based on an improved ecbbs algorithm

Abstract

The application provides a multi-AGV continuous task path planning method and system based on an improved ECBS algorithm. The method first decomposes the tasks of multiple AGVs into subtasks and plans the paths through an improved enhanced conflict resolution search algorithm. The improved algorithm dynamically estimates the cost of nodes using an online learning strategy and optimizes the resolution of path conflicts through a conflict priority mechanism and geometric symmetry. In addition, a heuristic weight graph is constructed to calculate path dependency, thereby improving planning efficiency and path quality. The system has real-time dynamic adjustment function and can adjust the AGV path according to environmental changes during task execution, adapting to complex scheduling requirements in multi-AGV continuous task scenarios. The application improves the task execution efficiency and path planning stability of the multi-AGV system and is suitable for intelligent warehousing, unmanned driving and logistics transportation fields.

Description

Technical Field

[0001] This application relates to the field of automated guided vehicles (AGVs) technology, and in particular to a method and system for continuous task path planning for multiple AGVs based on an improved ECBS algorithm. Background Technology

[0002] With the rapid development of the warehousing and logistics industry, automation technology has played a crucial role in improving warehousing efficiency and reducing labor costs. Among these technologies, AGVs, as an important component of intelligent warehousing and logistics systems, have gradually gained widespread application in warehouse transportation. The main task of AGVs is to automatically transport goods within the warehouse according to predetermined routes, and their efficient operation relies on accurate path planning technology. Therefore, the AGV path planning problem has become a key focus of research and application in warehousing and logistics automation.

[0003] The core of AGV path planning technology is to find the optimal path from the starting point to the destination for the AGV, while ensuring that the path avoids obstacles and conflicts with other vehicles during execution, and guarantees the efficient operation of the AGV. AGV path planning problems typically involve multiple objectives, such as shortest path, minimum time, and minimum energy consumption. These objectives require the path planning algorithm to generate travel routes quickly and accurately in dynamic and changing warehousing environments, while adapting to unexpected situations such as road congestion and equipment failure.

[0004] Traditional path planning methods are primarily based on graph search algorithms, such as the A* algorithm and Dijkstra's algorithm. These algorithms can search for the shortest path in a known environment and accelerate the search process through heuristic functions. The A* algorithm is one of the commonly used classic algorithms. By combining path length with heuristic values, it can quickly find a solution while ensuring the shortest path. However, the A* algorithm has high computational complexity when dealing with dynamic environments or large-scale warehousing environments, making it difficult to handle application scenarios with high real-time requirements.

[0005] In recent years, intelligent optimization algorithms and machine learning techniques have been increasingly incorporated into AGV path planning. Heuristic and metaheuristic algorithms, such as genetic algorithms, particle swarm optimization, and ant colony optimization, can search for globally optimal solutions under complex, multi-objective conditions. Compared to traditional graph search algorithms, intelligent optimization algorithms possess stronger global search capabilities, especially in path planning problems, avoiding getting trapped in local optima and ensuring the flexibility and adaptability of path planning. Furthermore, machine learning-based techniques (such as deep learning and reinforcement learning) are increasingly being applied to AGV autonomous navigation and path planning. These techniques, by learning motion patterns in complex environments, can plan reasonable paths for AGVs in dynamic and unknown environments.

[0006] However, in practical applications, AGV path planning algorithms are broadly categorized into traditional algorithms and population-based intelligent optimization algorithms. Traditional path planning methods, such as graph search algorithms (A* and Dijkstra), suffer from high computational complexity, particularly in obstacle-rich environments where path search efficiency is low. They also struggle to adapt to dynamic environmental changes, requiring path recalculation when obstacles or AGV positions change. Furthermore, they are ineffective at handling path conflicts and coordination issues in multi-AGV systems. In complex environments, population-based intelligent optimization algorithms, such as reinforcement learning and neural networks, offer certain advantages, adapting to dynamic environments to some extent. However, they require long training times, substantial data and computational resources, and the learning process is unstable, easily getting trapped in local optima and failing to guarantee a globally optimal path. In practical applications, AGV path planning technology faces numerous challenges. For example, warehousing environments are highly dynamic and uncertain; factors such as cargo accumulation and interference from other vehicles can lead to path congestion, thus affecting AGV operating efficiency. To address these challenges, researchers have introduced multi-vehicle collaborative scheduling and dynamic obstacle avoidance strategies, enabling multiple AGVs to work collaboratively in the same environment and dynamically adjust their travel paths through communication mechanisms to avoid collisions. Summary of the Invention

[0007] This application provides a multi-AGV continuous task path planning method and system based on an improved ECBS algorithm, which can be used to solve the technical problems of high computational complexity in multi-AGV systems that cannot well handle path conflicts and coordination issues.

[0008] A multi-AGV continuous task path planning method based on an improved ECBS algorithm, the method comprising the following steps:

[0009] Step 1: Flatten the task list of each AGV; before flattening, the task list is represented as a three-dimensional list, with the second dimension being the task and the third dimension being the target point; after flattening, the task list is represented as a two-dimensional list, with the second dimension being the target point; that is, all tasks of each AGV are regarded as an ordered list of a series of target points; finally, the task list of each AGV is converted into a stack.

[0010] Step 2: Iterate through all AGVs. Each time, pop a target point from the AGV's task stack as the AGV's endpoint in this path planning, and the starting point is the previous target point or the initial position. After iteration, you will get a set of corresponding starting and ending point lists.

[0011] Step 3: Preprocessing Start and End Points: For AGVs with the same end point, only one end point is kept unchanged, while the rest are modified. There are two modification methods: the first is to change it to a waypoint on the path from the start point to the end point; the second is to change it to the corresponding start point. After the modification, the end points that are now unreachable in this planning phase are pushed back onto the top of the corresponding AGV's task stack to ensure no end point is missed. Remove start-end point pairs with the same start and end point from the start and end point list, as they do not require planning and only need to wait (or occupy) the corresponding time step, which is related to the path planning result. Taking the second floor of a certain scenario as an example, since there are only two raw material upstairs areas connected to the workstations by only one path, and there are 20 AGVs in the raw material area, nearly 10 AGVs will simultaneously head to these two upstairs areas, easily leading to a deadlock that is difficult to resolve on the only connecting path. Therefore, the second modification strategy is more suitable. The preprocessed start and end point list is as follows: Figure 3 As shown, the key is the AGV number (starting from 0). There are a total of 20+16 AGVs in the first and second floors of the scene, with numbers ranging from 0 to 35.

[0012] Step 4: Path planning; Obtain path planning results based on ECBS; ECBS is a conflict-based planning algorithm that significantly reduces search events by searching for suboptimal solutions. It is suitable for path planning problems with large maps (over 100x100) or a large number of agents (over 50). Since the current scenarios generally involve large maps, and Scenario 1 has an excessive number of AGVs, the ECBS algorithm is used in all scenarios. Some path planning results are shown below. Figure 4 As shown, the coordinate values ​​have undergone linear processing; if the planning fails but there is a avoidance point, the deadlock resolution algorithm is called; if the planning fails but there is no avoidance point, the program ends with a failure and re-enters step 1 for task planning.

[0013] Step 5: Path Truncation: To ensure timely connection to the next task and prevent AGVs that complete their tasks early from waiting unnecessarily, the path planning results need to be truncated. Calculate the minimum path list length for all AGVs, which is the time step required for the fastest AGV to complete the task. To fully consider the loading and unloading time of the AGVs at the target point (i.e., the workstation), truncation can be performed at this minimum value or any time step forward. If any time step forward is ignored, truncation is performed at the minimum value. The truncated time step is the time step for this path planning. Finally, the truncated path is appended to the path list of the corresponding AGV.

[0014] Step 6: Task Connection: Update the starting point and maintain the ending point; after the path is truncated, for AGVs that have not reached the ending point (due to the path list being truncated midway), push the ending point back onto the top of the corresponding task stack; for all AGVs, take the last position of the truncated path list as their new starting point.

[0015] Step 7: Repeat steps 2 to 6 to complete the path planning for all tasks and output the path list for each AGV.

[0016] Furthermore, the ECBS algorithm employs a heuristic optimization strategy, using a heuristic weight graph to globally optimize the paths of multiple AGVs, thereby reducing path conflicts and time consumption during AGV task execution.

[0017] Furthermore, the method estimates the path cost of low-level nodes based on historical planning data and updates the cost function in real time during the planning process;

[0018] The cost function is as follows:

[0019]

[0020] Where, p i w(p) is the position of agent i at time t, where w(p) is the position of agent i at time t. i (t) is the weight of that position (in most cases, it may be 1, indicating that the cost of each step is equal);

[0021] The definition of conflict cost is:

[0022] C conflict (C)=λ·|C|

[0023] Where C is the set of conflicts that occur, |C| represents the number of conflicts, and λ is a moderating factor used to adjust the impact of conflict severity on total cost;

[0024] In ECBS, the total cost is a weighted sum of path cost and conflict cost; assuming the path cost for each agent is C. path (p i Then the total cost of the multi-agent system is expressed as:

[0025]

[0026] Where N is the number of agents, C represents all conflicts, and C0 is the number of agents. total It is the total cost of the system.

[0027] Furthermore, based on the number of conflicts caused by AGVs, the conflicts between AGVs are divided into fatal conflicts, minor conflicts, and non-fatal conflicts. Fatal conflicts are resolved first to ensure the continuity of tasks and execution efficiency.

[0028] Furthermore, the method also includes: Step 8, if a new task is added or other AGV paths change, the AGV path planning is adjusted in real time based on the monitored environmental changes, including the following steps:

[0029] Step 81: Calculate the number of deadlocks or conflicts for all AGVs, and reorder the AGVs in descending order of the number of conflicts, with the aim of prioritizing the AGV with the most conflicts to go to the avoidance point.

[0030] Step 82: Iterate through all the reordered AGVs. For each AGV, estimate the distance to all avoidance points using the Manhattan distance, and sort the avoidance points in ascending order of distance.

[0031] In the Manhattan distance formula, assuming there are two points (x1, y1) and (x2, y2), the Manhattan distance d can be expressed as:

[0032] d = |x1 - x2| + |y1 - y2|

[0033] Here, |x1-x2| is the distance in the horizontal direction, and |y1-y2| is the distance in the vertical direction.

[0034] Step 83: Using A * The algorithm iterates through all avoidance points, A * The algorithm is a heuristic for graph search, aiming to find the shortest path from a starting point to a target point. It combines a greedy strategy with Dijkstra's algorithm, optimizing path selection by evaluating the cost of each node during the search process; A * The algorithm uses the following cost function to evaluate the priority of each node:

[0035] f(n) = g(n) + h(n)

[0036] Where: g(n) is the actual cost from the starting point to the current node n, i.e., the path length; h(n) is the estimated cost from the current node n to the target node, i.e., the heuristic function; the A* algorithm continuously expands the node with the smallest f(n) value until the target node is found, ensuring that the path is optimal. The A* algorithm is used to solve the path from the AGV to the avoidance point. At this point, the positions of other AGVs are set as impassable, meaning that the AGV occupies the corresponding position; if the avoidance point is unreachable, the iteration continues until the first reachable avoidance point with the shortest distance is found; if all avoidance points are unreachable, the process returns to step 82 to iterate on the next AGV.

[0037] Step 84: Move the AGV currently in the iteration from Step 82 to the avoidance point found in Step 83. The path has also been solved. During this period, other AGVs wait in place.

[0038] Step 85: Attempt to plan paths for other AGVs. At this point, the avoidance point in Step 83 is impassable (it is occupied). If the planning is successful, the deadlock is successfully resolved, and during this period, the AGV at the avoidance point will wait. If the planning fails, return to Step 83 to iterate to the next avoidance point.

[0039] Step 86: After the deadlock is successfully resolved, continue to plan subsequent tasks based on the path planning algorithm; among them, the AGV at the avoidance point pushes its destination onto the top of the corresponding task stack (that is, the AGV does not complete the task in this planning, but is handed over to the next planning to complete it automatically, so as to avoid the other AGVs from being trapped in a long waiting time); return the path planning result of this deadlock resolution and enter step 85 of the path planning to connect the task stage and subsequent tasks.

[0040] This application also provides a multi-AGV continuous task path planning system based on an improved ECBS algorithm, the system comprising:

[0041] M1, Task Decomposition Module, is used to decompose the tasks of multiple AGVs into a continuous sequence of subtasks;

[0042] M2, the path planning module, plans a path for each AGV based on the improved ECBS algorithm;

[0043] M3, Conflict Detection and Resolution Module, is used to detect and resolve path conflicts between multiple AGVs;

[0044] M4, the dynamic adjustment module, is used to adjust the AGV's path according to real-time environmental changes during task execution.

[0045] This invention provides a multi-AGV continuous task path planning method and system based on an improved ECBS algorithm. The proposed improved ECBS algorithm focuses on handling the path planning problem when multiple AGVs are performing continuous tasks. This method not only retains the efficiency of the traditional ECBS algorithm in conflict detection and resolution, but also further improves the overall operating efficiency and robustness of AGVs by optimizing task allocation and dynamic obstacle avoidance strategies. In practical applications, this system can quickly generate reasonable paths, reduce task conflicts, and optimize vehicle scheduling when dealing with complex and dynamic environments in warehouses. Simultaneously, the system also incorporates a multi-vehicle cooperative mechanism to optimize energy consumption and rationally schedule resources while ensuring efficient AGV operation. Attached Figure Description

[0046] Figure 1 This is a flowchart of the AGV path planning method of the present invention;

[0047] Figure 2 This is a schematic diagram illustrating conflict detection during the path planning process of this invention.

[0048] Figure 3 This is a schematic diagram of scenario one provided in an embodiment of this application;

[0049] Figure 4 A path planning flowchart provided for an embodiment of this application. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0051] This invention treats the task list of each AGV as a stack structure, sequentially popping destination nodes from the tasks, and designating each path node in the task list as a destination where the AGV needs to stop, such as an operating station or charging station. For each AGV, an initial path to the corresponding destination is first generated independently based on its initial conditions, ensuring the optimality and safety of path planning, and conflicts are detected in real time during the path planning process.

[0052] The generated initial path undergoes conflict detection to determine if multiple AGVs collide at the same time point. If a path conflict is detected, candidate solutions are generated using a binary branch structure. Within each branch, constraints are imposed on the AGVs involved in the conflict to prevent them from avoiding the conflict area, and the path is replanned.

[0053] A focus search strategy based on suboptimal search is adopted, using a suboptimal coefficient as a constraint to quickly generate candidate paths that meet the shortest path condition, while ensuring that the path cost is within an acceptable range. Once the AGV completes the path for the current task node, the next destination is popped from the stack, and the above path planning steps are repeated until the task list is empty.

[0054] The present application is further illustrated below with reference to two embodiments.

[0055] Example 1

[0056] In multi-AGV task scenarios, each AGV typically needs to complete multiple consecutive transportation tasks within different time periods. This invention first decomposes each AGV's task into multiple sub-tasks in chronological order. Each sub-task includes a starting position and a target position, representing the process by which the AGV needs to move from the starting point to the target point.

[0057] After receiving task information from multiple AGVs, the system first parses the task sequence of each AGV based on the task decomposition module and transforms it into multiple sub-tasks, each sub-task corresponding to an independent path planning requirement.

[0058] Then, the system initializes the path planning process, including:

[0059] S1. Assign an initial path to each AGV based on the complexity of the task. The initial path is the shortest path for each task.

[0060] S2. Initialize global constraints, such as collision avoidance and safe distances between AGVs;

[0061] S3. Initialize the parameters of the improved ECBS algorithm, such as the weights of the heuristic function and the tolerance of suboptimal solutions.

[0062] By decomposing the task and planning the initial path, the system provides basic data input for subsequent conflict detection and resolution.

[0063] Example 2

[0064] In practical applications, the working environment of AGVs is often dynamically changing. For example, a new transportation task may be assigned to an AGV before an existing task is completed. To cope with this dynamic environment, this invention provides a real-time path adjustment mechanism. The steps are as follows:

[0065] S1. Dynamic Task Access: When a new task enters the system, the system first checks the task status of all AGVs. For AGVs that are close to completing their current task, the system will directly assign them a new task starting point and perform path planning. For AGVs that are currently executing, the system will determine whether their current path is affected by the new task and adjust the path as needed to avoid conflicts.

[0066] S2. Real-time Path Update: The improved ECBS algorithm supports real-time path updates. When a new conflict or environmental change is detected, the system will partially update the currently planned path without having to replan the paths for all AGVs from scratch. By adjusting the local planning for specific conflict areas, the system can quickly respond to dynamic changes in the environment, ensuring that the AGVs can continue to perform their tasks smoothly.

[0067] S3. AGV Path Priority Adjustment: The system adjusts path planning based on the urgency and priority of AGV tasks. For example, for important AGV tasks, the system can assign a higher priority path, while other AGVs need to adjust their paths without affecting the important tasks. This priority adjustment mechanism enables multi-task, multi-AGV systems to operate efficiently even when facing urgent tasks.

[0068] The embodiments described above do not constitute a limitation on the scope of protection of this application.

Claims

1. A multi-AGV continuous task path planning method based on an improved ECBS algorithm, characterized in that, The multi-AGV continuous task path planning method based on the improved ECBS algorithm includes the following steps: Step 1: Flatten the task list of each AGV; before flattening, the task list is represented as a three-dimensional list, with the second dimension being the task and the third dimension being the target point; after flattening, the task list is represented as a two-dimensional list, with the second dimension being the target point; that is, all tasks of each AGV are regarded as an ordered list of a series of target points; finally, the task list of each AGV is converted into a stack. Step 2: Iterate through all AGVs. Each time, pop a target point from the AGV's task stack as the AGV's endpoint in this path planning, and the starting point is the previous target point or the initial position. After iteration, you will get a set of corresponding starting and ending point lists. Step 3: Preprocessing start and end points; Step 4: Path planning: Obtain the path planning result based on ECBS; the coordinate values ​​have been linearized; if the planning fails and there is a avoidance point, call the deadlock resolution algorithm; if the planning fails and there is no avoidance point, the program ends with failure and re-enters Step 1 for task planning. Step 5: Path truncation. Calculate the minimum length of the path list for all AGVs, which is the time step required for the AGV to complete the task fastest. Truncation can be performed at this minimum or any time step forward. If any time step forward is ignored, truncation is performed at the minimum. The time step at the truncation point is the time step for this path planning. Finally, the truncated path is added to the path list of the corresponding AGV. Step 6: Task connection, update the starting point and maintain the ending point; Step 7: Repeat steps 2 to 6 to complete the path planning for all tasks and output the path list for each AGV.

2. The method according to claim 1, characterized in that, The improved ECBS algorithm adopts a heuristic optimization strategy, using a heuristic weight graph to globally optimize the paths of multiple AGVs.

3. The method according to claim 1, characterized in that, The method estimates the path cost of low-level nodes based on historical planning data and updates the cost function in real time during the planning process. The cost function is as follows: Where, p i w(p) is the position of agent i at time t, where w(p) is the position of agent i at time t. i (t) is the weight at that position; The definition of conflict cost is: C conflict (C)=λ·|C| Where C is the set of conflicts that occur, |C| represents the number of conflicts, and λ is a moderating factor used to adjust the impact of conflict severity on total cost; In ECBS, the total cost is a weighted sum of path cost and conflict cost; assuming the path cost for each agent is C. path (p i Then the total cost of the multi-agent system is expressed as: Where N is the number of agents, C represents all conflicts, and C0 is the number of agents. total It is the total cost of the system.

4. The method according to claim 1, characterized in that, The method categorizes conflicts between AGVs into fatal conflicts, minor conflicts, and non-fatal conflicts based on the number of conflicts caused by the AGVs, and resolves fatal conflicts first.

5. The method according to claim 1, characterized in that, The method further includes: step 8, if a new task is added or other AGV paths change, the path planning of the AGV is adjusted in real time based on the monitored environmental changes.

6. The method according to claim 5, characterized in that, Step 8: If new tasks are added or other AGV paths change, the AGV path planning is adjusted in real time based on the monitored environmental changes, including the following steps: Step 81: Calculate the number of deadlocks or conflicts for all AGVs, and reorder the AGVs in descending order of the number of conflicts; Step 82: Iterate through all the reordered AGVs. For each AGV, estimate the distance to all avoidance points using the Manhattan distance, and sort the avoidance points in ascending order of distance. Step 83: Using A * The algorithm iterates through all avoidance points, optimizing path selection by evaluating the cost of each node during the search process; A * The algorithm uses the following cost function to evaluate the priority of each node: f(n) = g(n) + h(n) Where: g(n) is the actual cost from the starting point to the current node n, i.e., the path length; h(n) is the estimated cost from the current node n to the target node, i.e., the heuristic function; the A* algorithm continuously expands the node with the smallest f(n) value until the target node is found; the A* algorithm solves the path from the AGV to the avoidance point, at which point the positions of other AGVs are set as impassable, which means that the AGV occupies the corresponding position; if the avoidance point is unreachable, continue iterating until the first reachable avoidance point with the shortest distance is found; if all avoidance points are unreachable, return to step 82 to iterate the next AGV; Step 84: Move the AGV currently in the iteration from Step 82 to the avoidance point found in Step 83. The path has also been solved. During this period, other AGVs wait in place. Step 85: Attempt to plan paths for other AGVs. At this point, the avoidance point in Step 83 is impassable. If the planning is successful, the deadlock is resolved, and the AGV at the avoidance point will wait indefinitely during this period. If the planning fails, return to Step 83 to iterate to the next avoidance point. Step 86: After the deadlock is successfully resolved, continue to plan subsequent tasks based on the path planning algorithm; among them, the AGV at the avoidance point pushes its destination onto the top of the corresponding task stack; return the path planning result of this deadlock resolution and enter step 85 of the path planning to connect the task stage and subsequent tasks.

Images