Multi-robot task balanced allocation method under energy constraint based on segmented anchoring

Abstract

The invention discloses a multi-robot task balanced distribution method under energy constraint based on segmented anchoring, which comprises the following steps of: firstly, distinguishing robot separation by marking path information with 0, 1 and-1, returning when the load is full or the electric quantity is insufficient, and generating an initial solution by hybrid coding of task number and completion proportion recorded by split information, and performing initial repair; in iteration, conventional population evolution is combined with probability to screen non-dominated solutions, and a charging interruption path is optimized through a sequence anchoring and balancing mechanism of segmented optimization, iterative anchoring and residual workload balancing; and finally, finishing time is finely adjusted by a re-balancing mechanism based on proportional splitting, and a non-dominated solution set is output. According to the method, the opportunity of converting passive charging constraint into active optimization is achieved, and the complex problem of multi-robot task allocation under energy limitation is systematically solved.

Description

Technical Field

[0001] This invention belongs to the field of intelligent agriculture and multi-robot scheduling technology, specifically relating to a multi-robot task balancing allocation method based on segmented anchoring under energy constraints. Background Technology

[0002] In smart agriculture scenarios, such as orchard harvesting, introducing multi-robot systems to replace manual labor is a key development direction. However, planning efficient operational schemes for robot teams is an extremely complex problem. This not only requires optimizing two conflicting objectives simultaneously—minimizing the team's overall completion time and total transportation costs—but also strictly adhering to the robots' own physical constraints, such as limited load-bearing capacity and battery capacity.

[0003] Existing technologies have significant shortcomings in dealing with such problems, especially in addressing energy constraints. When a robot has to return to the warehouse to replace its battery due to insufficient power, this forced charging disrupts its original work plan. Charging not only consumes time but also causes the robot's load to be reset to zero. This sudden change in state can have a cascading effect on the planning of all subsequent tasks, making the originally carefully designed path plan instantly no longer optimal or even infeasible.

[0004] Traditional multi-robot task allocation methods lack effective mechanisms to address the "planning disruption" problem caused by energy constraints. They typically cannot systematically repair and reconstruct disrupted task paths, resulting in inefficient solutions in practical applications.

[0005] Therefore, designing a method that can proactively respond to the dynamic effects of charging behavior and efficiently and systematically reconstruct and balance the disturbed path is a technical challenge that urgently needs to be solved in the current field. Summary of the Invention

[0006] To address the aforementioned issues, this invention proposes a multi-robot task balancing allocation method based on segmented anchoring under energy constraints. This method employs a novel anchoring and reconstruction paradigm, using path breakpoints caused by charging behavior as optimization anchor points. Around these anchor points, the disturbed path segments are systematically repaired, optimized, and balanced, thereby efficiently obtaining a task allocation scheme that balances completion time and transportation cost while satisfying energy constraints.

[0007] The technical solution adopted in this invention is: a multi-robot task balancing allocation method based on segmented anchoring under energy constraints, comprising the following steps:

[0008] S1, Initialization and Preliminary Repair: An initial solution population including path information and task splitting information is randomly generated using a hybrid encoding method for solutions; for each initial solution, the sequence anchoring and balancing mechanism is invoked for the first time, and a local search method is used to comprehensively repair and optimize the unreasonable path structure caused by energy constraints;

[0009] S2, Iterative Optimization: Before the preset termination condition is met, perform the following operations in a loop:

[0010] S21, Conventional Population Evolution: Based on a multi-objective evolutionary algorithm framework, the population is subjected to conventional iteration, and a new generation of solutions is generated through crossover and mutation operators;

[0011] S22, Elite Solution Deep Optimization: Select non-dominated solutions from the population with a preset probability, and perform sequence anchoring and balancing mechanisms on the non-dominated solutions to achieve deep optimization and reconstruction of the path structure;

[0012] S3, Final Equilibrium and Output: Near the end of the algorithm, a proportional split-based reequilibrium mechanism is executed on the current optimal non-dominated solution set to fine-tune the completion time of each robot; then the optimized non-dominated solution set is output.

[0013] Furthermore, in S1, the hybrid encoding method includes path information encoding and split information encoding. The path information encoding uses an integer sequence to represent the global task execution order and defines three special markers to distinguish the reasons for the robot returning to the warehouse: "0" is a robot separator used to distinguish the tasks of different robots; "1" indicates that the robot returns to the warehouse because it is full; "-1" indicates that the robot returns to the warehouse because it is low on power. The split information encoding uses a list or table to record the complete information of the split tasks. The information includes at least the task number and the completion ratio of the task on different robot paths.

[0014] Furthermore, in S22, the sequence anchoring and equalization mechanism is used to solve the path interruption problem caused by the charging triggering of the "-1" marker. The specific steps are as follows:

[0015] S221, Path Segmentation Optimization: The complete task sequence of a single robot is decomposed into multiple independent path segments according to the "-1" marker; each path segment is treated as a whole, and the macro execution order of the segments is adjusted through the sequence optimization algorithm to optimize the position of segments with high time consumption or energy consumption; then, local search optimization is performed on the task sequence inside the path segment that is interrupted due to charging.

[0016] S222, Segmented Anchoring: Perform iterative "anchoring-reconstruction" on the path sequence optimized in S221: First, locate the first "-1" marker in the path sequence, define the part from the path start point to the marker as an anchored segment and lock it; then, treat the remaining tasks after the marker as new subproblems and reinitialize them using the method in step S1 to generate a new path sequence; repeat S221 on the new path sequence until all tasks are planned and no new charging behavior is triggered; finally, concatenate all the sequentially generated "anchored segments" in order to form a complete path that satisfies the energy constraint;

[0017] S223, Balanced Remaining Workload: Collect the remaining tasks of all robots after their last charge and build a common task pool; re-plan the tasks in the task pool into new efficient path segments; based on the working time of each robot's anchored portion, allocate the new path segments to each robot according to the principle of filling the time difference, so as to minimize the team's maximum completion time.

[0018] Furthermore, in S3, the rebalancing mechanism based on proportional splitting specifically includes the following steps:

[0019] S31, Identify the bottleneck: Determine the bottleneck robot with the longest completion time in the current solution, as well as the last non-charging path segment executed by that robot, and record it as the donation path segment;

[0020] S32, Calculate the split ratio: Calculate the ideal average completion time of all robots, and determine the time margin of each non-bottleneck robot based on the average completion time; calculate the proportion of the task that each non-bottleneck robot should share from the donation path segment according to the time margin ratio of each non-bottleneck robot.

[0021] S33, Perform splitting and redistribution: Split the tasks in the donation path segment according to the calculated task volume ratio, and add the split tasks to the end of the task sequence of each non-bottleneck robot to balance the completion time of each robot.

[0022] Furthermore, the sequence optimization algorithm in S221 is the 2-opt sequence optimization algorithm.

[0023] The beneficial effects of this invention are:

[0024] 1. This invention innovatively proposes a segmented anchoring and balancing mechanism, which provides a systematic solution to the interruption-reconstruction problem of multi-robot path planning under energy constraints, and can transform passive charging behavior into an opportunity for active optimization;

[0025] 2. This invention decomposes the complex reconstruction process into three steps: macroscopic path segmentation optimization, iterative microscopic segmentation anchoring, and global residual workload balancing, thereby achieving multi-level collaborative optimization of disturbed paths from coarse to fine and from local to global.

[0026] 3. This invention introduces a rebalancing mechanism based on proportional splitting, which can refine the optimal solution, further squeeze the optimization potential of completion time, and improve the balance of the final solution.

[0027] 4. This invention encapsulates complex repair logic into an independent optimization module and calls it in a probabilistic manner. Without significantly increasing the complexity of conventional iterations, it effectively improves the evolutionary quality of the population and achieves a balance between solution efficiency and solution quality. Attached Figure Description

[0028] Figure 1 This is a flowchart illustrating the overall framework of the present invention;

[0029] Figure 2 This is a schematic diagram of the hybrid encoding structure of the solution in this invention. Detailed Implementation

[0030] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained based on these drawings without creative effort. To facilitate understanding of the present invention, the present invention will be described in more detail below with reference to the accompanying drawings and specific embodiments.

[0031] like Figure 1 As shown, this invention is a multi-robot task balancing allocation method based on segmented anchoring under energy constraints, comprising the following steps:

[0032] S1, Initialization and Preliminary Repair: An initial solution population including path information and task splitting information is randomly generated using a hybrid encoding method of solutions; then, for each initial solution, the "sequence anchoring and balancing mechanism" is called for the first time to perform a comprehensive repair and optimization of the unreasonable structure caused by energy constraints using local search methods such as 2-opt.

[0033] like Figure 2As shown, the hybrid encoding method in step S1 includes path information encoding and split information encoding. The path information encoding uses an integer sequence to represent the global task execution order, and uses three special markers to represent different reasons for returning to the warehouse: "0" is used as a robot separator to distinguish the tasks of different robots; "1" indicates that the return to the warehouse is due to full load; "-1" indicates that the return to the warehouse is due to insufficient power. The split information encoding uses a list or table to record the detailed information of the split tasks, including the task number and its completion ratio on different paths.

[0034] S2, Iterative Optimization: Before the termination condition is met, perform the following operations in a loop:

[0035] S21, Conventional Population Evolution: The existing multi-objective evolutionary algorithm framework is used to perform conventional iterations on the population, and new generation solutions are generated through operators such as crossover and mutation.

[0036] S22, Elite Solution Deep Optimization: Based on the regular iteration, a non-dominated solution in the population is selected with a preset probability to perform a "sequence anchoring and equilibrium mechanism" once to deeply optimize and reconstruct its path structure.

[0037] The "sequence anchoring and equalization mechanism" in step S22 aims to systematically reconstruct the path interrupted by charging (i.e., the "-1" marker), and specifically includes the following three steps:

[0038] S221, Path Segment Optimization Step: This step first decomposes a robot's complete task sequence into multiple independent path segments using the "-1" marker. Then, treating each path segment as a whole, the macroscopic execution order of these segments is adjusted using sequence optimization algorithms such as 2-opt, aiming to place time-consuming or energy-intensive path segments in more advantageous positions. Finally, for the task sequence within the specific path segment interrupted by charging, a local search optimization is performed again.

[0039] S222, Segmented Anchoring Step: This step iteratively "anchors and reconstructs" the path sequence optimized in S221. The specific process is as follows: First, find the first "-1" marker in the path sequence, define the portion from the path start point to this marker as an "anchored segment," and lock it. Then, treat all remaining tasks after this marker as a new, smaller subproblem, reinitialize them, and generate a new path sequence. Next, perform the path segment optimization of S221 again on this new sequence. Repeat the above process until all tasks are planned and no longer trigger new charging behaviors. Finally, concatenate all the sequentially generated "anchored segments" in order to form a structurally sound and energy-constrained complete path.

[0040] S223, Remaining Work Balancing Step: After all robot paths have undergone anchoring and refactoring in S222, this step aims to perform a final global load balancing. It collects all "remaining tasks" from each robot's last charge, forming a common task pool. Then, all tasks in this pool are treated as an independent planning problem, refactored into a series of new, efficient path segments. Finally, based on the work duration of each robot's anchored portion, and adhering to the principle of "filling the time gap," these newly generated path segments are greedily allocated to each robot, thereby minimizing the final maximum team completion time.

[0041] S3, Final Equilibrium and Output: As the algorithm is about to end, a "proportional split-based reequilibrium mechanism" is executed on the current optimal non-dominated solution set to perform a final fine-tuning of the completion time of each robot; finally, the optimized non-dominated solution set is output.

[0042] The "proportional split-based rebalancing mechanism" implemented in S3 aims to fine-tune the optimal solution at the end of the algorithm. The specific steps are as follows:

[0043] S31, Identify the bottleneck: Find the robot with the longest completion time in the current solution (bottleneck robot) and the last non-charging path segment it executed (donation path segment).

[0044] S32, Calculate the split ratio: Calculate the ideal average completion time for all robots, and based on this, calculate the "time margin" that each non-bottleneck robot can still handle. Based on the time margin of each robot, calculate the proportion of the task workload that each robot should share from the "donation path segment".

[0045] S33, Perform splitting and redistribution: Based on the calculated ratio, split the tasks in the "donation path segment" and add them to the end of the task sequences of other robots, thereby accurately "smoothing out peaks and filling valleys" to make the completion time of all robots more consistent.

[0046] To further illustrate the superiority of this invention in solving multi-objective multi-robot task allocation, Table 1 presents the results obtained by this invention and some excellent algorithms for solving multi-robot task allocation, namely Multi-objective Discrete Artificial Bee Colony (MODABC), Hybrid ant colony optimization-artificial bee colony algorithm (HACO), and Non-dominated Sorting Genetic Algorithm-Ⅱ (NSGA-Ⅱ), on the generated orchard multi-robot task allocation test set.

[0047] Table 1 Comparison of the optimal number of times the dataset experiments were conducted

[0048] Methods\Test Set 50×50 60×60 70×70 MODABC 0 0 0 HACO 0 1 0 NSGA-II 0 0 0 This invention 10 9 10

[0049] The examples present experimental results on a generated test set of multi-objective, multi-robot task allocation in an orchard. These problems vary in difficulty, with greenhouse sizes ranging from 50×50 to 70×70 square meters, containing 320, 500, and 720 fruit trees (task loads), respectively. The fruit tree maturity rate is set to 0.8 for each scenario, and the number of available robots is uniformly 4. Therefore, each combination of fruit tree number and maturity represents a different test problem to evaluate the algorithm's task allocation performance under different work scenarios and task difficulties. The experiment termination condition is set to a maximum CPU runtime of 0.5 × task load, and a population size of 30. To avoid the influence of randomness on the experimental results, each algorithm is run 10 times for each test problem. Hypervolume is used as the evaluation metric. Table 1 shows the number of problems in which each algorithm achieves optimal results in each test problem scenario. The comparison shows that the proposed method demonstrates superior optimization performance in solving multi-objective, multi-robot task allocation with power constraints. Specifically, regardless of the test problem scenario, the number of times the optimal result is obtained is greater than the number of times other algorithms obtain the optimal result.

[0050] In summary, this invention, through an innovative "anchoring-reconstruction-equilibrium" optimization paradigm, can transform passive charging constraints into opportunities for proactive optimization, systematically solving the complex problem of multi-robot task allocation under energy constraints, and providing decision-makers with a series of ideal compromise solutions.

[0051] This invention encapsulates complex path repair and reconstruction logic into an independent optimization mechanism and invokes it probabilistically rather than executing it in every iteration. This effectively controls computational overhead while guiding the population to evolve toward a better structure, providing a valuable reference for solving other complex scheduling problems affected by interruption constraints.

[0052] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for balanced task allocation among multiple robots under energy constraints based on segmented anchoring, characterized in that, Includes the following steps: S1, Initialization and Preliminary Repair: An initial solution population including path information and task splitting information is randomly generated using a hybrid encoding method for solutions; for each initial solution, the sequence anchoring and balancing mechanism is invoked for the first time, and local search methods are used to comprehensively repair and optimize the unreasonable path structure caused by energy constraints; S2, Iterative Optimization: Before the preset termination condition is met, perform the following operations in a loop: S21, Conventional Population Evolution: Based on a multi-objective evolutionary algorithm framework, the population is subjected to conventional iteration, and a new generation of solutions is generated through crossover and mutation operators; S22, Elite Solution Deep Optimization: Select non-dominated solutions from the population with a preset probability, and perform sequence anchoring and balancing mechanisms on the non-dominated solutions to achieve deep optimization and reconstruction of the path structure; S3, Final Equilibrium and Output: Near the end of the algorithm, a proportional split-based reequilibrium mechanism is executed on the current optimal non-dominated solution set to finely adjust the completion time of each robot. Then output the optimized non-dominated solution set.

2. The method for balanced allocation of multi-robot tasks under energy constraints based on segmented anchoring as described in claim 1, characterized in that, In S1, the hybrid encoding method includes path information encoding and split information encoding. The path information encoding uses an integer sequence to represent the global task execution order and defines three special markers to distinguish the reasons for the robot returning to the warehouse: "0" is a robot separator used to distinguish the tasks of different robots; "1" indicates that the robot returns to the warehouse because it is full; "-1" indicates that the robot returns to the warehouse because it is low on power. The split information encoding uses a list or table to record the complete information of the split tasks. The information includes at least the task number and the completion ratio of the task on different robot paths.

3. The method for balanced allocation of multi-robot tasks under energy constraints based on segmented anchoring as described in claim 1, characterized in that, In S22, the sequence anchoring and equalization mechanism is used to solve the path interruption problem caused by the charging triggering of the "-1" marker. The specific steps are as follows: S221, Path Segmentation Optimization: The complete task sequence of a single robot is decomposed into multiple independent path segments according to the "-1" marker; each path segment is treated as a whole, and the macro execution order of the segments is adjusted through the sequence optimization algorithm to optimize the position of segments with high time consumption or energy consumption; then, local search optimization is performed on the task sequence inside the path segment that is interrupted due to charging. S222, Segmented Anchoring: Perform iterative "anchoring-reconstruction" on the path sequence optimized in S221: First, locate the first "-1" marker in the path sequence, define the part from the path start point to the marker as an anchored segment and lock it; then, treat the remaining tasks after the marker as new subproblems and reinitialize them using the method in step S1 to generate a new path sequence; repeat S221 on the new path sequence until all tasks are planned and no new charging behavior is triggered; finally, concatenate all the sequentially generated anchored segments in order to form a complete path that satisfies the energy constraint. S223, Balanced Remaining Workload: Collect the remaining tasks of all robots after their last charge and build a common task pool; re-plan the tasks in the task pool into new efficient path segments; based on the working time of each robot's anchored portion, allocate the new path segments to each robot according to the principle of filling the time difference, so as to minimize the team's maximum completion time.

4. The method for balanced allocation of multi-robot tasks under energy constraints based on segmented anchoring as described in claim 1, characterized in that, In S3, the rebalancing mechanism based on proportional splitting specifically includes the following steps: S31, Identify the bottleneck: Determine the bottleneck robot with the longest completion time in the current solution, as well as the last non-charging path segment executed by that robot, and record it as the donation path segment; S32, Calculate the split ratio: Calculate the ideal average completion time of all robots, and determine the time margin of each non-bottleneck robot based on the average completion time; calculate the proportion of the task that each non-bottleneck robot should share from the donation path segment according to the time margin ratio of each non-bottleneck robot. S33, Perform splitting and redistribution: Split the tasks in the donation path segment according to the calculated task volume ratio, and add the split tasks to the end of the task sequence of each non-bottleneck robot to balance the completion time of each robot.

5. A method for balanced allocation of multi-robot tasks under energy constraints based on segmented anchoring, as described in claim 3, is characterized in that... The sequence optimization algorithm in S221 is the 2-opt sequence optimization algorithm.

Images