Multi-robot task cooperation path planning method, device and product
Patent Information
- Application Number
- CN202511798096.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-02
- Publication Date
- 2026-03-06
Smart Images
Figure CN121612296A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of automatic control technology, and in particular to a method, apparatus and product for multi-robot collaborative path planning. Background Technology
[0002] The need for multi-robot task collaboration and path planning has permeated all sectors of national economic and social development. In intelligent manufacturing, multi-robot collaborative handling and assembly significantly improve production efficiency and flexibility. In warehousing and logistics, a large number of Automated Guided Vehicles (AGVs) need to operate in parallel within narrow passages, requiring path planning algorithms to generate collision-free paths in real time to optimize sorting efficiency. In environmental monitoring and disaster relief scenarios, drones or ground robot swarms need to autonomously plan coverage paths to ensure comprehensive detection of target areas. In the field of autonomous driving, unmanned vehicle fleets need to reduce wind resistance and save energy through collaborative path planning, while also responding to unexpected road conditions. These application scenarios pose multiple challenges to multi-robot path planning, including real-time performance, safety, and efficiency.
[0003] The core challenges of multi-robot task collaboration and path planning include the exponential growth of computational complexity with the number of robots, conflict avoidance, dynamic environmental adaptability, and communication and resource constraints. As the number of robots increases, the search space for task allocation and path planning expands exponentially, making it difficult for traditional algorithms to find feasible solutions within a finite timeframe. Furthermore, potential collisions between robots, the need for task collaboration, and the emergence of dynamic obstacles all require algorithms to adjust paths in real time. In environments with limited communication or resources, designing distributed algorithms to achieve local collaboration while ensuring global task optimization has become a pressing technical challenge.
[0004] C. Mahulea et al., in their paper "Robot Planning based on Boolean Specifications using Petri Net Models," proposed an automated method for planning paths in multi-robot systems. This method uses a Petri net with output to model the motion capabilities and regions of interest of the multi-robot system. The proposed task specification is transformed into linear constraints on a set of binary variables; the robot's mobility is expressed as linear constraints based on the Petri net states; and the values of the binary variables are associated with the Petri net through linear inequalities. By solving an integer linear programming problem, the solution yields the robot trajectory that satisfies the task requirements.
[0005] Existing technologies for path planning in multi-robot systems do not take into account task collaboration between robots, resulting in poor path execution efficiency and scenario adaptability; they are prone to getting trapped in local optima, leading to suboptimal planning results; in addition, multi-robot path planning using Petri nets involves a large amount of computation and has poor real-time performance. Summary of the Invention
[0006] The purpose of this invention is to provide a multi-robot task collaborative path planning method, apparatus, and product to address all or part of the problems mentioned above, thereby solving the problem of efficient, highly adaptable, and highly efficient planning of collaborative task paths for multi-robot systems.
[0007] The technical solution adopted in this invention is as follows: A multi-robot collaborative path planning method, comprising: S1. Obtain the environmental map and task requirements for multi-robot operations; the environmental map contains multiple regional units, and the task execution time cost of each unit is fixed; the task requirements indicate the prohibited units in the environmental map, the task units in which each robot performs tasks independently, the collaborative units in which multiple robots need to perform tasks simultaneously, and the destination units that each robot finally reaches. S2. Under the constraints of the task requirements, with the goal of minimizing the total time for all robots to complete all tasks, the path for each robot to perform tasks is iteratively optimized by simulated annealing. The optimization result when the deadline is reached is used as the planned path for each robot.
[0008] In another aspect, the present invention also provides a multi-robot task collaborative path planning device, including a processor and a storage medium, wherein the storage medium stores computer instructions, and the processor executes the computer instructions to perform the above-described multi-robot task collaborative path planning method.
[0009] In addition, the present invention also provides a computer program product, including a computer program, which, when run by a processor, executes the above-described multi-robot task cooperative path planning method.
[0010] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: This application's multi-robot task collaborative path planning method addresses high-dimensional combined constraint scenarios where multi-robot systems perform collaborative tasks. It utilizes simulated annealing to plan and solve multi-robot paths under constraints, achieving high computational efficiency and strong real-time performance. By incorporating a temperature decay strategy to balance global exploration and local search, it avoids getting trapped in local optima, exhibiting greater adaptability to various scenarios and capable of solving path planning problems in more complex task scenarios, thus improving the concurrency of multi-task execution. Furthermore, using the total task execution time as the optimization objective improves the overall efficiency of multi-robot task execution. This application preprocesses the environment map using a shortest path algorithm, calculating the shortest path between any two units (excluding prohibited units) for direct use in the optimization process, further enhancing optimization efficiency. Moreover, this application's multi-robot task collaborative path planning method requires minimal adjustments when adding robots or tasks; only task requirements need to be added, with almost no impact on computational load and efficiency, demonstrating strong scalability. Attached Figure Description
[0011] The present invention will be described by way of example and with reference to the accompanying drawings, wherein: Figure 1 This is a flowchart of a multi-robot task collaborative path planning method in one embodiment.
[0012] Figure 2 This is a schematic diagram of an individual code in one embodiment.
[0013] Figure 3 , Figure 4 , Figure 5 The diagrams illustrate the implementation of random swapping, the 2-Opt algorithm, and the re-insertion method, respectively.
[0014] Figure 6 This is a schematic diagram of the environment map in a small-scale scenario embodiment.
[0015] Figure 7 yes Figure 6 The planning result diagram in the embodiment.
[0016] Figure 8 This is a diagram showing the planning results in a large-scale scenario implementation. Detailed Implementation
[0017] All features disclosed in this specification, or all steps in all disclosed methods or processes, may be combined in any way, except for mutually exclusive features and / or steps.
[0018] Any feature disclosed in this specification (including any appended claims and abstract) may be replaced by other equivalent or similar features, unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is merely one example of a series of equivalent or similar features.
[0019] To address the issues of poor real-time performance, unsuitability for task collaboration scenarios, and susceptibility to local optima in current multi-robot system path planning that does not consider task collaboration, this application proposes a multi-robot task collaboration path planning method, apparatus, and product, aiming to solve the problem of high real-time performance and high accuracy in path planning under multi-robot system task collaboration scenarios.
[0020] like Figure 1 As shown, the multi-robot task cooperative path planning method proposed in this application includes the following process: S1. Obtain the environmental map and task requirements for multi-robot operations.
[0021] The environment map contains multiple regional units, and the task execution time and cost of each unit are fixed.
[0022] The environment map is the area where all robots operate. The environment map is divided into multiple units, each unit corresponding to a regional task. The task execution time cost of the task in each unit is the same for different robots. That is, the task execution time cost of the same unit is the same regardless of which robot performs it.
[0023] The task requirements indicate prohibited units in the environment map, task units where each robot performs tasks independently, collaborative units where multiple robots need to perform tasks simultaneously, and destination units where each robot ultimately reaches.
[0024] The task requirements clearly define all tasks to be performed by all participating robots (including independent and collaborative tasks), prohibited areas, and the final unit to be reached after completing all tasks. Collaborative tasks require all robots to arrive at the corresponding collaborative unit before execution; that is, if a robot arrives at a collaborative unit first, it must wait for the last robot to arrive before executing the corresponding collaborative task.
[0025] To facilitate efficient machine processing and improve planning efficiency, as an optional implementation method, the method for obtaining task requirements includes: The task requirements include prohibited units, task units where each robot can independently perform tasks, collaborative units where multiple robots need to perform tasks simultaneously, and the final destination units that each robot must reach. Task requirements are typically issued in the form of work orders, and the relevant task data is obtained by parsing the corresponding tables within the work order.
[0026] Furthermore, the prohibition unit, the task unit in which each robot performs its task independently, the collaborative unit in which multiple robots perform their tasks simultaneously, and the destination unit that each robot ultimately reaches are represented in Boolean logic specification format.
[0027] The above methods enable formal modeling of various tasks, laying the foundation for subsequent multi-robot task allocation and collaborative path planning.
[0028] For example, definition The transformed Boolean logic specification consists of four sub-specifications: Equation (1): ; In the formula, sub-specification Y represents the logical requirements for the task unit where each robot independently performs its task, expressed as: Equation (2): ; In equation (2), m represents the number of subtasks. This represents the i-th subtask. Indicates corresponding to The set of units for the i-th task to be executed. This represents a subset of cells containing tasks to be performed. Indicates the logical AND. Indicates the logical "OR"; That is, the Boolean constraint for the i-th subtask, indicating that it can be performed in the cell set. any subset of units in To carry out the mission.
[0029] by Represents a set of units for independent tasks, where Let C represent the set of units for the i-th task to be executed, where each unit is represented by c, and the environment map is represented by C. .
[0030] For example, for the first robot, its corresponding sub-specification This represents the first task unit where the robot independently performs a task, containing Boolean constraints for the first and second subtasks. The first set of task units to be executed contains only the first subset of units. The first subtask unit set Includes the first unit subset This indicates that the first robot needs to access the first subset of units. One of the units in the set; the second set of task units to be executed is a subset of the second unit. Or the third unit subset Boolean constraints for the second subtask Includes the second unit subset and the third unit subset Any subset of units in the array indicates that the first robot still needs to visit the second subset of units. One of the units, or the third subset of units. One of the units in it.
[0031] Sub-specification T represents the logical requirements for a collaborative unit where multiple robots perform tasks simultaneously. It requires multiple robots (of various types or in large quantities) to complete the same task concurrently within the same unit. The collaborative task only begins execution when all required robots have arrived at the collaborative unit. This is represented as: Equation (3): ; In the formula, n represents the number of sub-cooperative units. Let j represent the set of the j-th sub-cooperative units. Indicates corresponding to The j-th unit set, Let j be the set of the j-th sub-cooperative units, and let j be the set of its corresponding units. any subset of units in ;r represents any robot required for the collaborative task. Denotes the set of the j-th sub-cooperative units The required set of robots, Denotes the set of the j-th sub-cooperative units Require All robots must be accessed.
[0032] by Represents the set of units for collaborative tasks, where This represents the set corresponding to the j-th sub-cooperative unit. The j-th unit set.
[0033] For example, for a collaborative unit in an environment map, its corresponding sub-specification: ; That is, there is only one sub-cooperative unit set, which contains only one subset of units. The collaborative task was completed by the first and second robots working together.
[0034] Sub-specification A represents the logical requirement for the prohibited unit, expressed as: Equation (4): ; In the formula, This represents the set of areas where access is prohibited, and the set of prohibited units is represented as... . express any subset of units in Access is prohibited to all.
[0035] For example, the sub-specification A of the environment map C is: ; That is, for a subset of units All cells are prohibited from access; a subset of cells must be avoided during path planning. All units in.
[0036] Sub-specification U represents the logical requirements of the final destination unit reached by each robot. It is represented as: Equation (5): ; In the formula, q represents the number of sub-destination unit sets. Denotes the set of the k-th destination units. Indicates corresponding to The k-th unit set, Let the set of the k-th destination units be denoted by and let be the set of units corresponding to it. any subset of units in .
[0037] by Denotes the set of destination units, where This represents the set of the k-th unit.
[0038] For example, for the first robot, its sub-specification U is: ; For the second robot, its sub-specification U is: ; That is, the final destination unit of the first robot is In any unit, the final destination unit of the second robot is Any unit in it.
[0039] Using the above method, all task requirements in a multi-robot system's collaborative path planning for a single task are constrained in the same Boolean logic format to facilitate unified processing by the machines.
[0040] S2. Under the constraints of the task requirements, with the goal of minimizing the total time for all robots to complete all tasks, the path for each robot to perform tasks is iteratively optimized by simulated annealing. The optimization result when the deadline is reached is used as the planned path for each robot.
[0041] The above task requirements constrain the prohibited units, task units, cooperative units, and destination units of each robot. Under these constraints, each robot needs to visit the task unit and cooperative unit at least once and reach the destination unit while avoiding the prohibited units, so as to minimize the total time to complete all tasks.
[0042] As an optional implementation, before iteratively optimizing the path for each robot to perform its task, the shortest path between every two reachable units (i.e., non-prohibited units in the environment map) in the environment map that avoids prohibited units is calculated, so as to provide an environment topology that can be directly invoked for the iterative optimization of the path for each robot to perform its task.
[0043] The so-called environmental topology is a constraint that requires the robot to travel along the shortest path between two units. For example, for units c1 and c2, the shortest path is c1c5c2. There is also a reachable path c1c6c2, but this path is longer than c1c5c2. Therefore, it is agreed that when the robot visits c2 from unit c1, only the path c1c5c2 is considered, and there are no other alternatives. This can save a lot of path attempts.
[0044] In one alternative implementation, the shortest path between every two reachable cells that avoids prohibited cells is calculated using the Dijkstra algorithm.
[0045] The specific calculation method for the shortest path is shown in Algorithm 1:
[0046] The above method enables the preprocessing of the environmental map, providing basic environmental topology constraints for the subsequent path optimization process. It also provides directly calculable path length data for use, thereby eliminating unnecessary trial schemes during the optimization process, reducing data computation, and improving planning efficiency.
[0047] Under the aforementioned task conditions and environmental topology constraints, the path for each robot to perform the task can be iteratively optimized using simulated annealing. The process includes: S21. Based on the task requirements and environmental topology, initialize the paths for each robot to perform the task.
[0048] Under environmental topology constraints, the corresponding initialization path can be mapped by initializing the order in which each robot traverses at least one round of task units and cooperative units and finally lands in the destination unit.
[0049] In one specific implementation, such as Figure 2As shown, individual coding is used to plan the paths of each robot.
[0050] Each robot's possible task execution sequence (i.e., unit access order) is encoded as an individual Ind, which is divided into two parts: a task region part IY and a final region part IU. The individual length is specified by Boolean logic. The number of robots, k, determines this.
[0051] The task region IY is defined as follows: Equation (6): ; IY is a one-dimensional dictionary containing k keys and m positive integers, where m, as mentioned earlier, represents the number of subtasks. This refers to the k-th robot. Element This represents the identifier of the j-th task for the i-th robot, indicating the task. . , where represents the total number of tasks for the i-th robot.
[0052] The final region portion of IU is represented as: ; IU is also a one-dimensional dictionary containing k keys and k positive integers. , where is the identifier of the destination unit for the i-th robot, if and This represents the destination unit. .
[0053] The specific implementation method of individual encoding is shown in Algorithm 2:
[0054] For example, the individual obtained through individual coding is: This indicates that the path of the first robot is c3c6c8, and the destination unit is c9; the path of the second robot is c1c7c8, and the destination unit is c5.
[0055] S22. Calculate the total time for each robot to complete all tasks.
[0056] Based on the unit distance time cost (a fixed value) and the task execution time cost of each unit, the time consumed by each robot in executing the task according to the path can be calculated. The longest time consumed by all robots is the total time to complete all tasks.
[0057] The total time for each robot to complete all tasks is calculated as shown in Algorithm 3:
[0058] S23. Determine whether the total time has reached the deadline. If yes, end the optimization process. Otherwise, under the constraints of the task requirements, dynamically adjust the path of at least one robot based on the current paths of each robot, and return to step S22. Repeat this process until the deadline is reached.
[0059] Each time the path of the multi-robot system is adjusted, the total time to complete all tasks needs to be calculated and compared with the previous total time to determine whether it is better, thereby continuously optimizing the path of the multi-robot system.
[0060] Simulated annealing can be used to solve the task allocation problem in large-scale multi-robot systems. This algorithm primarily aims to minimize the total task completion time, thereby effectively improving the system's task concurrency capabilities.
[0061] As an optional implementation method, under task requirements constraints, a method for dynamically adjusting the path of at least one robot based on the current paths of all robots includes: Under the constraints of the task requirements, the path of at least one robot is dynamically adjusted using random swapping, the 2-Opt algorithm, or the re-insertion method. After dynamic adjustment, a new individual is obtained. .
[0062] This application provides three methods for dynamically adjusting paths: (1) Random exchange Random swapping refers to obtaining a new path by randomly exchanging the positions of any two task units within the original path. Taking the path planning implemented through individual encoding as an example, random swapping would then involve randomly swapping task regions. The new task area is obtained by identifying the positions of two task unit identifiers of a certain robot. ,like Figure 3 As shown.
[0063] (2) 2-Opt algorithm The 2-Opt algorithm involves randomly selecting any two tasks in the original path and reversing the order of all tasks between those two tasks. Taking individual encoding as an example, this is achieved by partially reversing the order of tasks within the task region. In a sequence of task identifiers for a certain robot, two task identifiers are randomly selected and reversed to obtain a new task region. ,like Figure 4 As shown.
[0064] (3) Reinsertion method Reinsertion refers to the method of re-insertion for a portion of the task region. Randomly select a task and another task Replace, and obtain the new task area section. For the final area portion Randomly select a destination unit And use another destination unit (if any). Replace to obtain the new final area portion The dynamic transformation process of the re-insertion method is as follows: Figure 5 As shown.
[0065] The individual mentioned above For example, new individuals are obtained by applying the above three dynamic adjustment methods respectively: (1) Random exchange The new individual obtained is: .
[0066] (2) 2-Opt algorithm The new individual obtained is: .
[0067] (3) Reinsertion method The new individual obtained is: .
[0068] A new individual is obtained by dynamically adjusting the path of at least one robot. The total time for each robot to complete all tasks after the dynamic adjustment is calculated and compared with the current total time for all robots to complete all tasks. It is then determined whether the acceptance condition is met. If so, the dynamically adjusted paths of all robots are accepted; otherwise, the path of at least one robot in the current paths of all robots is dynamically adjusted again.
[0069] Based on the concept of simulated annealing, the acceptance criteria for the dynamically adjusted path are determined, with two possibilities: absolute acceptance and probabilistic acceptance. Specifically, in one optional implementation, it is determined whether the total time for all robots to complete all tasks after dynamic adjustment is less than the current total time for all robots to complete all tasks. If so, the acceptance criteria are met (i.e., absolute acceptance); otherwise, the acceptance criteria are considered met based on the acceptance probability of the current iteration (i.e., probabilistic acceptance). This acceptance probability is determined based on the ratio of the total time difference before and after dynamic adjustment to the temperature control coefficient of the current iteration.
[0070] Specifically, since the task execution time cost of each regional unit is fixed, the factor affecting the total time to complete all tasks is the total distance traveled. Define the objective function. This represents the total distance traveled, calculated from the individual Ind. If ,in For example, the adjustments mentioned above That is, the dynamically adjusted new individual will accept the adjusted path and... As the starting state for the next iteration; otherwise, even based on The calculated objective function value is greater than the total computation time. Still based on the probability of acceptance Accept the adjusted path, where This represents the temperature control coefficient for the current iteration round.
[0071] To avoid getting trapped in local optima during the optimization process and to balance global exploration with local search, as an optional implementation method, the temperature control coefficient for the current iteration is determined by the temperature parameter for the current iteration. The Boltzmann constant k is determined. The temperature parameters for the next iteration are then determined. From the cooling coefficient and the temperature parameters of the current iteration round The calculations were performed. The initial temperature parameters were set as follows: Specifically, .
[0072] In the implementation method that controls the iteration through a temperature control coefficient, a preset final temperature parameter is used. The temperature parameter is used as the cutoff condition for the iteration, i.e., when the temperature parameter drops to a certain value. When the time is reached, the optimization process terminates, and the final optimization result is output as the planned path for each robot.
[0073] Algorithm 4 illustrates the method for iteratively optimizing the path of a multi-robot system using simulated annealing.
[0074] In this application embodiment, the performance of the proposed multi-robot task collaborative path planning method was also verified in application scenarios of different scales.
[0075] (1) Small-scale scenarios like Figure 6 As shown, the defined environment map contains 10 regional units, represented as follows: . This indicates a multi-robot system containing a first robot and a second robot. Initially, the first robot is located in cell c1, and the second robot is located in cell c5. This represents the initial cell containing all robots, where Let the initial cell, i.e. the initial position, be the location of the i-th robot. Then we have: .
[0076] Region of Interest Set Defined as: ,in , , , , , , , .
[0077] The task requirements after Boolean logic canonical transformation are: .
[0078] Wherein, the sub-standard Y is: ; .
[0079] That is, the first robot should access cell c2, and at least one of cells c3, c5 or c6; the second robot should access cell c1 and cell c7.
[0080] Sub-specification T is: .
[0081] This indicates that the first and second robots should complete the collaborative task at unit c8.
[0082] Sub-specification A is: .
[0083] This indicates that no robot can access unit c10.
[0084] Sub-standard U is: ; .
[0085] The destination cell where the first robot finally stops is cell c5 or cell c9; the destination cell where the second robot finally stops is cell c7.
[0086] Based on the above task requirements, we have: ; , ; ; ; , .
[0087] The distance between any two cells is represented by the adjacency matrix M of the environment map. M is a 10×10 symmetric matrix, represented as follows: ; In the formula, Indicates two units The average distance traveled between units. When they are not adjacent, it is considered It is infinitely large.
[0088] Assume the average distance between adjacent units is: ; The results of the minimum distance calculation are shown in Table 1.
[0089] Table 1 Shortest Path Set
[0090] in, Representation unit and unit The shortest distance between, Representation unit and unit The shortest path between.
[0091] Since the collaborative task is one that both the first and second robots need to perform, when considering the task of a single robot, the set of collaborative units can be incorporated into the set of task units required for each robot. That is... , .
[0092] Individual coding is performed based on task requirements. According to Algorithm 2, individual coding is performed on a multi-robot system containing the first and second robots. The coded individual representation is as follows: .
[0093] Assume the encoded individual is: .
[0094] The task execution order of each robot can be obtained by decoding this individual Ind: .
[0095] Based on the shortest path calculated from Table 1, the path planning scheme for the i-th robot can be determined. and the corresponding shortest distance ,in: .
[0096] The objective function value corresponding to individual Ind Defined as: Equation (7): ; In this embodiment, k=2.
[0097] Based on Table 1, the shortest paths and shortest distances for the first and second robots, obtained through decoding, are as follows: .
[0098] The total distance traveled is calculated from this. (units).
[0099] Assuming all robots travel at a speed of 1 unit, the distance a robot travels is the same as the time it takes to travel that distance. Two robots are engaged in a cooperative task. This means that the collaborative task must begin when both the first and second robots reach c8. The task execution time cost for each unit is defined as... Algorithm 3 calculates the time each robot takes to complete all tasks (including travel time and task execution time), thus accurately determining the total time required for all robots to complete all tasks. Algorithm 3 first initializes the starting unit, destination unit, and travel distance for each robot. Then, it progressively advances the state of each robot in a time-step loop: for moving robots, it reduces the remaining distance; upon reaching a task unit, it checks if it's a collaborative unit requiring mutual processing; if so, it waits for other robots to arrive; otherwise, it continues to the next task unit. By dynamically maintaining the path queue and collaborative task markers for each robot, it ultimately returns the total time for the complete execution process.
[0100] During the iteration process, different dynamic adjustment methods can be used to adjust individuals in different embodiments. For example, for an individual... The individuals adjusted using the random exchange method are: Individuals adjusted using the 2-Opt algorithm are: The individuals adjusted using the re-insertion method are: .
[0101] By continuously iterating and optimizing the paths of each robot, such as Figure 7 As shown, the final shortest path is: ; .
[0102] When the first robot leaves the unit Departure to Unit The time required is 9, and the second robot starts from... Departure Arrival The required time is 7 (<9), but due to This is a shared task for both the first and second robots, so the second robot needs to... The system waits for the first robot to arrive until 9:00 AM. Then, both robots begin working, consuming 1 unit of time, at which point the time is 10. Afterward, the first and second robots complete their respective tasks: the first robot starts from... Departure Arrival It takes 4 hours to complete the task, and 1 hour to complete it; from Departure Arrival It takes 3 hours to complete the task, and 1 hour to complete it; from Departure Arrival The first robot requires 0 time to complete the task, and 1 time is consumed to finish it; the total time consumed is t1 = 10 + 1 + 4 + 1 + 2 + 1 = 19. The second robot starts from... Departure Arrival It takes 1 minute to complete the task, and 5 minutes to finish it; from Departure Arrival It takes 10 seconds to complete the task, and 2 seconds to finish it; from Departure Arrival It takes 2 hours to complete the task, and 1 hour to complete the task; the total time t2 = 10 + 1 + 5 + 10 + 2 + 2 + 1 = 31. Therefore, the shortest time required for all robots to complete all tasks is 31.
[0103] (2) Large-scale scenarios In this scenario, assume the environment map contains 67 area units, i.e. The average distance between adjacent cells is a randomly generated integer between 1 and 10.
[0104] The multi-robot system consists of three robots: the initial units of the first robot, the second robot, and the third robot are c1, c10, and c67, respectively.
[0105] The set of regions of interest is ,in
[0106]
[0107]
[0108]
[0109] ; =[7,3,5,9,2,8,1,6,4,10,2,5,7,1,3,8,4,9,6,2,10,1,5,7,3,6,8,4,9,2,5,1,7,3,6,9,4,8,2,10,3,5,1,7,9,4,6,2,8,10,1,3,5,7,9,2,4,6,8,10,2,4,6,8,10,1,7].
[0110] Assuming Boolean logic specification In this context, the sub-standard Y is: ; ; .
[0111] Then we have: ; ; .
[0112] Assume the sub-specification T is: .
[0113] This indicates that the first and second robots are simultaneously performing a cooperative task at cell c8. .
[0114] Assume sub-specification A is: .
[0115] This indicates that the robot is prohibited from passing through the cell along the planned trajectory. , .
[0116] Assume the sub-standard U is: ; ; .
[0117] This indicates that the first robot should eventually remain in the unit. or The second robot should remain in the unit. The third robot should remain in the unit. . , .
[0118] Through iterative optimization, the optimal task allocation is finally obtained as follows: The first robot: ; The second robot: ; The third robot: .
[0119] The calculated total time is 131, and the planned path is as follows: Figure 8 As shown.
[0120] Experiments in both small-scale and large-scale scenarios demonstrate that the multi-robot task collaborative path planning method proposed in this application can improve the high concurrency of multi-robot systems, effectively reduce the total time for multi-robot systems to execute collaborative tasks, and improve task execution efficiency. Furthermore, the path planning time is extremely short in all scenarios, demonstrating high real-time planning performance.
[0121] Based on the ideas of this application, this application also proposes a multi-robot task collaborative path planning device, including a processor and a storage medium. The storage medium stores computer instructions, and the processor runs the computer instructions to execute the multi-robot task collaborative path planning method of the above embodiments.
[0122] In addition, this application also proposes a computer program product, including a computer program, which, when run by a processor, executes the multi-robot task cooperative path planning method described above.
[0123] This invention is not limited to the specific embodiments described above. The invention extends to any new feature or combination disclosed in this specification, as well as any new method or process step or combination disclosed herein.
Claims
1. A multi-robot task coordination path planning method, characterized in that, The method comprises the following steps: S1, obtaining an environment map of a multi-robot task and a task requirement; The environment map comprises a plurality of area units, and the task execution time cost of each unit is fixed; The task requirement indicates forbidden units in the environment map, task units for each robot to independently execute tasks, cooperative units for multiple robots to simultaneously execute tasks, and destination units for each robot to finally arrive at; S2, under the constraint of the task requirement, iteratively optimizing the paths of the robots for executing tasks to minimize the total time for all robots to complete all tasks, and taking the optimization result when the stop condition is reached as the planned path of each robot.
2. The multi-robot task coordination path planning method of claim 1, wherein, Before iteratively optimizing the paths of the robots for executing tasks, the shortest paths for avoiding the forbidden units between each two reachable units in the environment map are calculated, so as to provide an environment topology that can be directly called for the process of iteratively optimizing the paths of the robots for executing tasks.
3. The multi-robot task coordination path planning method of claim 1, wherein, The method for obtaining the task requirement comprises: Obtaining the forbidden units, the task units for each robot to independently execute tasks, the cooperative units for multiple robots to simultaneously execute tasks, and the destination units for each robot to finally arrive at involved in the task requirement; The forbidden units, the task units for each robot to independently execute tasks, the cooperative units for multiple robots to simultaneously execute tasks, and the destination units for each robot to finally arrive at are respectively expressed in a Boolean logic specification format.
4. The multi-robot task coordination path planning method of claim 2, wherein, The method for iteratively optimizing the paths of the robots for executing tasks by using the simulated annealing method comprises: S21, initializing the paths of the robots for executing tasks according to the task requirement and the environment topology; S22, calculating the total time for all robots to complete all tasks; S23, judging whether the total time reaches a stop condition, if yes, ending the optimization process, otherwise, dynamically adjusting the path of at least one robot on the basis of the current paths of the robots, and returning to step S22; This cycle is repeated until the stop condition is reached.
5. The multi-robot task coordination path planning method of claim 4, wherein, The method for dynamically adjusting the path of at least one robot on the basis of the current paths of the robots under the constraint of the task requirement comprises: Under the constraint of the task requirement, the path of at least one robot in the current paths of the robots is dynamically adjusted by using a random exchange method, a 2-Opt algorithm, or a reinsertion method; The total time for all robots to complete all tasks after the dynamic adjustment is calculated, and compared with the total time for all robots to complete all tasks before the dynamic adjustment, to judge whether an acceptance condition is met, if yes, the paths of the robots after the dynamic adjustment are accepted, otherwise, the path of at least one robot in the current paths of the robots is re-adjusted.
6. The multi-robot task coordination path planning method of claim 5, wherein, The method for judging whether the acceptance condition is met comprises: Judging whether the total time for all robots to complete all tasks after the dynamic adjustment is less than the total time for all robots to complete all tasks before the dynamic adjustment, if yes, the acceptance condition is met, otherwise, the acceptance condition is considered to be met based on an acceptance probability of the current iteration round, and the acceptance probability is determined based on the ratio of the difference between the total times before and after the dynamic adjustment to a temperature control coefficient of the current iteration round.
7. The multi-robot task coordination path planning method of claim 6, wherein, The temperature control coefficient of the current iteration round is determined by a temperature parameter of the current iteration round and a Boltzmann constant, and a temperature parameter of a next iteration round is calculated by a temperature reduction coefficient and the temperature parameter of the current iteration round. 8.The method of claim 6, wherein, The iteration termination condition is that the temperature parameter reaches a preset final temperature parameter. 9.A multi-robot task coordination path planning apparatus comprising a processor and a storage medium, characterized in that, The storage medium stores computer instructions, and the processor executes the computer instructions to perform the multi-robot task cooperative path planning method according to any one of claims 1-8.
10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to perform the multi-robot task cooperative path planning method according to any one of claims 1-8.