Multi-limb cooperative control method and system of bionic structure

Abstract

The invention provides a multi-limb cooperative control method and system of a bionic structure, and relates to the technical field of dynamic motion control, and the method comprises the steps: carrying out the sequential carding of to-be-executed tasks of a bionic robot, and building a to-be-executed task chain; performing cooperative control decision on the N limbs, and establishing a Q-dimensional robot control space; performing inspection optimization according to the robot control inspection channel, and establishing a robot control Q-dimensional candidate space; introducing a robot control variation function and a cooperative fitness analysis model to perform variation expansion optimization on the robot control Q-dimensional candidate space to obtain a robot control optimization block; and executing multi-limb cooperative control. Through the method and the device, the technical problem that the task execution efficiency of the bionic robot is low due to the fact that multi-limb cooperative control of the robot is difficult to adapt to complex and changeable tasks and environments in the prior art can be solved, the adaptability of a multi-limb cooperative control strategy in the complex environment is enhanced, and the task execution efficiency of the bionic robot is improved.

Description

Technical Field

[0001] This application relates to the technical field of dynamic motion control, and particularly to a multi-limb collaborative control method and system with a bionic structure. Background Art

[0002] A bionic robot is a robot that imitates the structure and functions of a living organism. Its core concept is to design a system that can simulate the movement and perception of a living organism to achieve efficient operation ability and adaptability. Multi-limb collaborative control involves the coordinated work of multiple robot limbs (such as robotic arms, mobile platforms, etc.), which can efficiently complete complex tasks and usually requires considering multiple factors such as the mechanical relationship, task requirements, and execution order between each limb. Currently, most multi-limb collaborative control methods rely on task presets and fixed control strategies, and usually use optimization algorithms to coordinate the collaboration of multiple limbs. When the task requirements change, the existing control methods usually need to recalculate or adjust the control strategy, resulting in a large computational overhead, thereby affecting real-time performance and efficiency. In addition, there are also problems of unbalanced collaboration between multiple limbs in the existing technology. Especially when facing complex and changing tasks and environments, problems such as uneven load and improper coordination may occur between limbs, resulting in task execution failure or low efficiency.

[0003] In summary, there is a technical problem in the existing technology that the task execution efficiency of bionic robots is low because it is difficult to adapt to complex and changing tasks and environments in the multi-limb collaborative control of robots. Summary of the Invention

[0004] The purpose of this application is to provide a multi-limb collaborative control method and system with a bionic structure to solve the technical problem in the existing technology that the task execution efficiency of bionic robots is low because it is difficult to adapt to complex and changing tasks and environments in the multi-limb collaborative control of robots.

[0005] In view of the above problems, this application provides a multi-limb collaborative control method and system with a bionic structure.

[0006] In a first aspect, the present application provides a multi-limb collaborative control method for a bionic structure, and the multi-limb collaborative control method for a bionic structure is implemented through a multi-limb collaborative control system for a bionic structure. Among them, the multi-limb collaborative control method for a bionic structure includes: sequentially sorting out the tasks to be executed by the bionic robot to establish a task chain to be executed. Among them, the bionic robot includes N limbs, and the task chain to be executed includes Q node tasks, and both N and Q are positive integers greater than 1; making a collaborative control decision on the N limbs according to the task chain to be executed to establish a Q-dimensional robot control space; performing inspection and optimization on the Q-dimensional robot control space according to a robot control inspection channel to establish a Q-dimensional candidate space for robot control; introducing a robot control mutation function and a collaborative fitness analysis model to perform mutation expansion optimization on the Q-dimensional candidate space for robot control to obtain a robot control optimization block; performing multi-limb collaborative control of the bionic robot according to the robot control optimization block.

[0007] In a second aspect, the present application further provides a multi-limb collaborative control system for a bionic structure, which is used to execute the multi-limb collaborative control method for a bionic structure as described in the first aspect. Among them, the multi-limb collaborative control system for a bionic structure includes: a task sequential sorting module, which is used to sequentially sort out the tasks to be executed by the bionic robot to establish a task chain to be executed. Among them, the bionic robot includes N limbs, and the task chain to be executed includes Q node tasks, and both N and Q are positive integers greater than 1; a collaborative control decision module, which is used to make a collaborative control decision on the N limbs according to the task chain to be executed to establish a Q-dimensional robot control space; an inspection and optimization module, which is used to perform inspection and optimization on the Q-dimensional robot control space according to a robot control inspection channel to establish a Q-dimensional candidate space for robot control; a mutation expansion optimization module, which is used to introduce a robot control mutation function and a collaborative fitness analysis model to perform mutation expansion optimization on the Q-dimensional candidate space for robot control to obtain a robot control optimization block; a collaborative control execution module, which is used to perform multi-limb collaborative control of the bionic robot according to the robot control optimization block.

[0008] One or more technical solutions provided in the present application have at least the following technical effects or advantages: By temporally arranging the tasks to be executed by the bionic robot and establishing a task chain to be executed, where the bionic robot includes N limbs, and the task chain to be executed includes Q node tasks, both N and Q are positive integers greater than 1; making a collaborative control decision for the N limbs according to the task chain to be executed, and establishing a Q-dimensional robot control space; performing inspection and optimization on the Q-dimensional robot control space according to the robot control inspection channel, and establishing a Q-dimensional candidate space for robot control; introducing a robot control mutation function and a collaborative fitness analysis model to perform mutation expansion optimization on the Q-dimensional candidate space for robot control, and obtaining an optimized block for robot control; performing multi-limb collaborative control of the bionic robot according to the optimized block for robot control. That is to say, by temporally arranging the tasks to be executed and establishing a task chain, making a collaborative control decision for each limb according to the task chain, and selecting the optimal control strategy through the robot control inspection channel, the robot control mutation function and the collaborative fitness analysis model for multi-limb collaborative control of the bionic robot, optimizing the collaborative actions between the limbs, enhancing the adaptability of the multi-limb collaborative control strategy in a complex environment, and improving the task execution efficiency of the bionic robot.

[0009] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of this application more obvious and understandable, the following specifically illustrates the specific embodiments of this application. It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of this application, nor is it used to limit the scope of this application. Other features of this application will become easily understood through the following description of the specification. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in this application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings described below are only exemplary. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0011] Figure 1 It is a schematic flowchart of a method for multi-limb collaborative control of a bionic structure in this application; Figure 2 It is a schematic structural diagram of a multi-limb collaborative control system of a bionic structure in this application.

[0012] Description of the reference numerals: Task temporal arrangement module 11, collaborative control decision module 12, inspection and optimization module 13, mutation expansion optimization module 14, collaborative control execution module 15. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0013] The present application provides a multi-limb collaborative control method and system with a bionic structure, which solves the technical problem in the prior art that the execution efficiency of bionic robots is low because it is difficult to adapt to complex and changeable tasks and environments in the multi-limb collaborative control of robots. By sequentially sorting the tasks to be executed and establishing a task chain, collaborative control decisions are made for each limb according to the task chain. Through the robot control test channel, the robot control mutation function, and the collaborative fitness analysis model, the optimal control strategy is selected for the multi-limb collaborative control of the bionic robot, optimizing the collaborative actions between the limbs, enhancing the adaptability of the multi-limb collaborative control strategy in complex environments, and improving the task execution efficiency of the bionic robot.

[0014] Next, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present application. Additionally, it should be noted that for the sake of description, only the parts related to the present application are shown in the drawings rather than all.

[0015] Embodiment 1, please refer to the attached Figure 1 , the present application provides a multi-limb collaborative control method with a bionic structure. Among them, the multi-limb collaborative control method with a bionic structure is applied to a multi-limb collaborative control system with a bionic structure. The multi-limb collaborative control method with a bionic structure specifically includes the following steps: S100: Sequentially sort the tasks to be executed by the bionic robot and establish a task chain to be executed. Among them, the bionic robot includes N limbs, and the task chain to be executed includes Q node tasks. Both N and Q are positive integers greater than 1.

[0016] Specifically, determine all the tasks to be executed by the bionic robot, and determine the nature, objectives, and execution conditions of each task. According to the importance and dependency of the tasks, sort the tasks in the execution order and establish a task chain to be executed containing Q node tasks. In the task chain, the input and output of each node task need to meet certain logical relationships to ensure that the execution of each task does not interfere with each other and can maximize the efficiency. The bionic robot includes N limbs, and each limb is responsible for a specific function or task.

[0017] A bionic robot is a robot that mimics the structure and functions of organisms, aiming to imitate the movement patterns, sensing capabilities, and control mechanisms of living beings. A bionic robot generally consists of multiple limbs, which work in coordination through a precise control system to perform complex tasks. In a robot, a limb refers to the part that performs movement and operation tasks, similar to the arms, legs, or other moving parts of an organism. A bionic robot can have multiple limbs, and each limb is responsible for a specific function or task. A to-be-executed task refers to a task that has not been executed yet but needs to be completed, usually including a series of operation steps such as grasping, transporting, and assembling. Each task may require the cooperation of multiple limbs to complete.

[0018] Task sequencing refers to arranging and organizing the to-be-executed tasks in a certain order to ensure that each task can be executed on time and in sequence. A task chain refers to an ordered sequence composed of a group of related tasks, and each task node depends on the execution result of the previous task. By sequencing and establishing a task chain for the to-be-executed tasks of the bionic robot, the execution order of each task is ensured to be reasonable, thereby reducing unnecessary waiting and resource conflicts and improving the overall execution efficiency of the bionic robot.

[0019] S200: Make a collaborative control decision for the N limbs according to the to-be-executed task chain, and establish a Q-dimensional robot control space.

[0020] Furthermore, S200 of this application includes: According to the to-be-executed task chain, extract the task of the q-th node, where 1 ≤ q ≤ Q; collect the scene information of the task of the q-th node to obtain the q-th task scene; use the q-th task scene as the collaborative control scene constraint and the task of the q-th node as the collaborative control target to make a control decision for the N limbs, and establish the q-th node robot control domain; add the q-th node robot control domain to the Q-dimensional robot control space.

[0021] Specifically, locate and extract the task of the q-th node from the established to-be-executed task chain. The task of the q-th node refers to the specific task of the q-th node in the task chain. For example, if the task chain includes a transportation operation, the task of the q-th node may be to grasp an object or place an object. According to the requirements of the task node, collect the relevant scene information of the task of the q-th node to obtain the q-th task scene, including the position and state of the object, the position of obstacles, environmental constraints, etc. The scene information can be collected through sensors (such as lidar, depth camera sensors, etc.) to obtain external environment data related to the task. The q-th task scene is the environmental information collected when executing the task of the q-th node, and the scene of each task node is related to the specific operation of that node.

[0022] According to the requirements of the q-th node task, the q-th task scenario is used as a constraint for collaborative control, that is, in the process of multi-limb collaborative control, physical or operational limitations related to the scenario must be observed. The q-th node task is used as the collaborative control target, that is, the goal that the robot needs to achieve in the collaborative control process. The collaborative control target of each task node is usually the operation to be completed at that node. According to the collaborative control scenario constraints and the collaborative control target, control decisions are made for N limbs to plan how to use these limbs to complete the q-th node task. Determine the specific action strategy of the robot under the q-th node task, including the coordination method and control method of each limb, and establish the control domain of the q-th node robot.

[0023] The control domain of the q-th node robot is all possible decision spaces for robot control when executing the q-th node task, including multiple robot control decisions corresponding to the q-th node task, that is, it includes all possible action, strategy, and decision combinations of the robot's limbs. Each operation mode may lead to different task execution results.

[0024] Add the control domain of the q-th node task to the Q-dimensional robot control space. The Q-dimensional robot control space is the set of control domains corresponding to all task nodes, including the Q node control domains corresponding to Q node tasks. The control decision domain of each task node represents the possible control strategies under that node. The Q-dimensional control space integrates the decision information of all nodes to form a complete robot control decision space. By analyzing the task chain to be executed by the bionic robot and making control decisions for each task, efficient collaborative control of multiple robot limbs can be achieved, and comprehensive management of the multi-limb collaborative control of the robot can be realized.

[0025] S300: According to the robot control inspection channel, inspect and optimize the Q-dimensional robot control space to establish a Q-dimensional candidate space for robot control.

[0026] Furthermore, S300 of this application includes: Traverse the control domain of the q-th node robot, extract the first robot control decision; according to the first robot control decision, simulate the control of the bionic robot to obtain the first simulated cooperation data set; input the first simulated cooperation data set into the robot control inspection channel to obtain the first robot control inspection result; when the first robot control inspection result is qualified, set the first robot control decision as the first candidate robot control solution, and add the first candidate robot control solution to the control candidate domain of the q-th node robot; when the first robot control inspection result is unqualified, eliminate the first robot control decision; continue to inspect and optimize the control domain of the q-th node robot according to the robot control inspection channel, construct the control candidate domain of the q-th node robot, and add the control candidate domain of the q-th node robot to the Q-dimensional candidate space of robot control.

[0027] Specifically, randomly select a node task in the Q-dimensional robot control space as the control domain of the q-th node robot. Traverse the control domain of the q-th node robot. Each control domain contains multiple possible control decisions, and randomly extract the first robot control decision from them. According to the first robot control decision, simulate the control of the bionic robot, simulate the execution process of the bionic robot under the first robot control decision, and obtain the first simulated cooperation data set, which includes the motion trajectory, grasping force, limb cooperation effect, etc. of the bionic robot. The first simulated cooperation data set refers to the actual execution results obtained by simulating the bionic robot after executing the first decision, and is used to evaluate the effect of the control strategy.

[0028] Input the first simulated cooperation data set into the robot control evaluation model in the robot control inspection channel for multi-dimensional index evaluation to obtain the first robot control evaluation result, including task completion accuracy, limb cooperation efficiency, and limb cooperation reliability. Then input the obtained first robot control evaluation result into the control inspection model, and judge whether the first robot control evaluation result meets the robot control inspection constraints through the control inspection operator. If it meets, the first robot control inspection result is qualified, otherwise it is unqualified.

[0029] When the inspection result of the first robot control is qualified, set the first robot control decision as the first candidate solution for robot control and add it to the robot control candidate domain of the q-th node. When the inspection result of the first robot control is unqualified, that is, the first robot control decision cannot meet the quality standards for task execution, such as the task accuracy, collaboration efficiency, or reliability not meeting the standards, automatically remove this control decision from the control decision space, that is, no longer continue to use this control decision to avoid its negative impact on the robot task execution. Continue to inspect and optimize other decisions in the robot control domain of the q-th node through the robot control inspection channel. That is to say, eliminate all unqualified decisions in the robot control domain of the q-th node, leave all qualified decisions, and construct a complete robot control candidate domain for the q-th node. The robot control candidate domain of the q-th node is all qualified decisions screened according to the inspection results during the inspection and optimization process and is the candidate control decisions available for selection.

[0030] Add the robot control candidate domain of the q-th node to the Q-dimensional candidate space for robot control. That is, perform the above steps for each node task to obtain Q robot control candidate domains corresponding to Q node tasks, which together constitute the Q-dimensional candidate space for robot control. The control candidate domain of each node task represents all optimized control decisions for that node, and the candidate domains of multiple nodes form the overall optimization space for robot control. By verifying the feasibility of each control decision, it is ensured that the robot can execute smoothly in the actual task, improving the success rate and control accuracy of the task, eliminating unqualified decision schemes, ensuring that the finally selected control scheme can adapt to different environmental and task conditions, avoiding unnecessary trial-and-error processes, and enhancing the overall execution efficiency.

[0031] Furthermore, the present application further includes the following steps: The robot control inspection channel includes a robot control evaluation model and a control inspection model; input the first simulated collaboration data set into the robot control evaluation model to obtain the first robot control evaluation result. Among them, the robot control evaluation model includes multi-dimensional evaluation indicators for robot control, and the multi-dimensional evaluation indicators for robot control include task completion accuracy, limb collaboration efficiency, and limb collaboration reliability; input the first robot control evaluation result into the control inspection model to output the first robot control inspection result. Among them, the control inspection model includes a control inspection operator, and the control inspection operator includes that if the first robot control evaluation result meets the robot control inspection constraints, the first robot control inspection result is qualified, and if the first robot control evaluation result does not meet the robot control inspection constraints, the first robot control inspection result is unqualified. The robot control inspection constraints include a predetermined task completion accuracy, a predetermined limb collaboration efficiency, and a predetermined limb collaboration reliability.

[0032] Specifically, the robot control inspection channel gradually screens and verifies whether the robot control decision meets the task requirements through the combination of the robot control evaluation model and the control inspection model. The robot control evaluation model is used to evaluate the execution effect of the robot control decision. According to multi-dimensional evaluation indicators, it measures the performance of the control decision and analyzes aspects such as the accuracy of the robot during task execution, the limb collaboration efficiency, and the limb collaboration reliability. The first simulated collaboration dataset is input into the robot control evaluation model for processing, and the indicators for each dimension are calculated according to the task requirements to obtain the first robot control evaluation result.

[0033] The robot control multi-dimensional evaluation indicators are multiple dimension indicators that measure the quality of the robot control decision, including task completion accuracy, limb collaboration efficiency, and limb collaboration reliability. Each indicator reflects different aspects of the robot during task execution. Task completion accuracy is the difference between the actual operation result and the predetermined target when the robot executes a task. The higher the accuracy, the closer the task completion effect is to the expected target. Limb collaboration efficiency is the efficiency of collaboration between multiple limbs of the robot, usually measured by coordination time, the total number of executed actions, etc. The higher the efficiency, the smoother the coordination between the robot's limbs and the faster the task is completed. Limb collaboration reliability is the ability of each limb of the robot to maintain stability and successfully complete the task during collaboration. The higher the reliability, the more consistently the robot can execute the task when performing the same task multiple times.

[0034] The task completion accuracy is obtained by calculating the deviation between the actual execution result of the robot and the target result. For example, assuming the initial position is (0,0), the target position is the coordinate (5,5), and the actual position is the coordinate (5.1,5.2), then the task completion accuracy is obtained by calculating the ratio of the distance between the target position and the actual position to the distance between the target position and the initial position, and then calculating the difference between 1 and the ratio, resulting in a task completion accuracy of 96.84%. The limb collaboration efficiency can be obtained by calculating the degree of cooperation of multiple limbs during task execution, such as the coordination time of arm movements and the number of executed actions. For example, assuming the total collaboration time of the left and right arms is 5 seconds, while the optimal collaboration time required by the task target is 6 seconds, then the collaboration efficiency is 120%, indicating that the task is completed in a shorter time than expected and the collaboration efficiency is high. The limb collaboration reliability can be obtained by simulating the execution of the same task multiple times and calculating the task success rate. For example, when the robot executes a task multiple times, the probability of successfully completing the task is 80% and the probability of failure is 20%, then the reliability is 80%.

[0035] Input the first robot control evaluation result into the control inspection model for final inspection. Compare the first robot control evaluation result with the robot control inspection constraints through the control inspection operator to determine whether the requirements are met, that is, whether the first robot control inspection result is qualified. The control inspection operator is a mathematical tool or algorithm used to determine whether a robot control decision is qualified. It inputs the control evaluation result and outputs whether it is qualified. The robot control inspection constraints are various requirements that the robot needs to meet when performing tasks, including predetermined standards such as task completion accuracy, limb collaboration efficiency, and limb collaboration reliability. For example, the task completion accuracy needs to reach or exceed 95%, the collaboration efficiency needs to reach or exceed 100%, and the collaboration reliability needs to reach or exceed 85%.

[0036] If the first control evaluation result meets the robot control inspection constraints, the first robot control inspection result is qualified; otherwise, if the first robot control evaluation result does not meet the robot control inspection constraints, the first robot control inspection result is unqualified. That is to say, when all three control indicators meet the predetermined standards, it indicates that this control decision is qualified. By combining the control evaluation model and the control inspection model, multi-dimensional analysis and verification of the control decision are carried out, and unqualified solutions are eliminated in a timely manner, so as to ensure stable task execution in different environments and conditions and avoid collaboration failures caused by unqualified decisions.

[0037] S400: Introduce the robot control mutation function and the collaboration fitness analysis model to perform mutation expansion optimization on the Q-dimensional candidate space of the robot control, and obtain the robot control optimization block.

[0038] Furthermore, S400 of this application includes: Mutate the Q-dimensional candidate space of the robot control according to the robot control mutation function to establish a Q-dimensional mutation space of the robot control; perform inspection optimization on the Q-dimensional mutation space of the robot control according to the robot control inspection channel to obtain a Q-dimensional mutation optimization space of the robot control; expand the Q-dimensional candidate space of the robot control according to the Q-dimensional mutation optimization space of the robot control to obtain a Q-dimensional optimization space of the robot control; perform maximization optimization of the collaboration fitness on the Q-dimensional optimization space of the robot control according to the collaboration fitness analysis model to obtain the robot control optimization results of Q nodes, where the collaboration fitness analysis model includes the task completion accuracy weight, the limb collaboration efficiency weight, and the limb collaboration reliability weight; store the robot control optimization results of the Q nodes both on-chain and off-chain to generate the robot control optimization block.

[0039] Specifically, the robot control mutation function is used to mutate the original Q-dimensional candidate space. By mutating the control schemes of Q task nodes, new control schemes are generated. Specifically, the mutation value of the control evaluation result corresponding to each control candidate scheme is evaluated through the robot control mutation function to obtain the corresponding candidate scheme mutation value coefficient. According to the candidate scheme mutation value coefficient, the number of mutations of the candidate scheme is determined, and the control candidate domain of Q task nodes is mutated, that is, the parameters of the control scheme are changed to obtain the robot control Q-dimensional mutation space. The specific process is described in detail in the subsequent corresponding dependent claim explanations. For the sake of simplicity of the specification, it will not be elaborated here.

[0040] The robot control Q-dimensional mutation space is inspected and optimized through the robot control inspection channel, similar to the process of establishing the Q-dimensional robot control space through the robot control inspection channel described above. For the sake of simplicity of the specification, only a brief introduction is given here and no detailed description is provided. Taking the robot control mutation space of the first node as an example, the first mutation decision is extracted from it. According to the first mutation decision, the bionic robot is simulated and controlled, and the data during the simulation process is recorded for evaluating the effect of the mutation decision. The simulation data is input into the robot control evaluation model in the robot control inspection channel to obtain the evaluation result of the first mutation decision, including task completion accuracy, limb cooperation efficiency, and limb cooperation reliability. Then, the evaluation result is input into the control inspection model in the robot control inspection channel to determine whether the evaluation result meets the robot control inspection constraints and whether it is qualified. The qualified schemes are retained and the unqualified schemes are eliminated, thereby obtaining the robot control mutation optimization space of the first node and inputting it into the robot control Q-dimensional mutation optimization space. For other node tasks, the above steps are all executed, thus constituting the robot control Q-dimensional mutation optimization space.

[0041] According to the robot control Q-dimensional mutation optimization space, the robot control Q-dimensional candidate space is expanded. That is to say, the optimized control schemes obtained through mutation and inspection are incorporated into the robot control Q-dimensional candidate space, expanding the scope of the candidate schemes to obtain a new Q-dimensional optimization space, namely the robot control Q-dimensional optimization space. The robot control Q-dimensional optimization space not only includes the schemes in the original robot control Q-dimensional candidate space but also all the control schemes after mutation and inspection.

[0042] Input the Q - dimensional optimization space of robot control into the collaborative fitness analysis model for fitness maximization optimization. By weighting factors such as task completion accuracy, limb collaboration efficiency, and limb collaboration reliability, a comprehensive fitness index is calculated to help select the optimal control scheme. The collaborative fitness analysis model includes the weights of task completion accuracy, limb collaboration efficiency, and limb collaboration reliability, which are used to represent the influence degree of each evaluation index on the final decision. By adjusting these weights, priority adjustment can be carried out in different tasks. In the robot control inspection channel, retrieve the evaluation results of each control scheme in the Q - dimensional optimization space of robot control, and perform weighted calculation with the weights of task completion accuracy, limb collaboration efficiency, and limb collaboration reliability to obtain the collaborative fitness of each scheme in the Q - node tasks.

[0043] Select the scheme with the maximum collaborative fitness among the Q - node tasks as the optimization result of robot control for the Q - node. Store the optimization result of robot control for the Q - node in the way of blockchain technology to generate a robot control optimization block. Store the most critical data on the blockchain. For larger amounts of data (such as detailed data of control schemes, evaluation indicators, etc.), they can be stored in an off - chain database. The data in the block is immutable, and all robot control schemes for executing this task will be recorded and can be queried and verified at any time.

[0044] On - chain storage means directly storing data in the blockchain network. The blockchain has the characteristics of immutability and decentralization, ensuring the security and reliability of data. The advantages of on - chain storage are data transparency, permanent storage, and public verifiability. Off - chain storage means storing data in a storage system outside the blockchain, such as a traditional database or a distributed storage system. Off - chain storage usually can provide higher storage capacity and faster data access, but compared with on - chain storage, its security and transparency are weaker. A robot control optimization block refers to a data block stored in the blockchain, which contains relevant data about the robot control optimization process, such as the control scheme for task execution, task completion accuracy, limb collaboration efficiency, reliability, and other optimization results.

[0045] Through mutation and optimization, continuously improve the control scheme, which can cope with complex and changeable tasks and environments, generate the most suitable control scheme for the current task, and improve the accuracy, efficiency, and reliability of task completion, especially showing better collaboration ability in multi - robot collaboration.

[0046] Furthermore, this application also includes the following steps: Connect the robot control inspection channel, and retrieve each robot control evaluation result corresponding to each robot control candidate scheme in the q-th node robot control candidate domain; perform mutation value evaluation on each robot control candidate scheme according to the each robot control evaluation result, and obtain the mutation value coefficient of each candidate scheme; input the mutation value coefficient of each candidate scheme into the robot control mutation function, and establish the q-th mutation constraint rule, wherein the q-th mutation constraint rule includes the mutation number of each candidate scheme corresponding to each robot control candidate scheme; mutate the q-th node robot control candidate domain according to the q-th mutation constraint rule, generate the q-th node robot control mutation domain, and add the q-th node robot control mutation domain to the robot control Q-dimensional mutation space.

[0047] Specifically, the robot control inspection channel is connected, and the robot control evaluation results corresponding to each robot control candidate scheme in the q-th node robot control candidate domain are retrieved, including the task completion accuracy, limb collaboration efficiency and limb collaboration reliability of each robot control candidate scheme. According to the robot control evaluation results, each robot control candidate scheme is evaluated for variation value, and the variation value coefficient of each candidate scheme is obtained. The variation value coefficient is a coefficient calculated by variation value evaluation, which characterizes the variation potential of each candidate control scheme and is used to determine the amplitude of variation. The higher the variation value coefficient, the greater the potential and necessity of variation of the scheme, which may have a greater impact on the effect of task execution. The robot control evaluation results are standardized and normalized, and different weights are assigned to each indicator according to the importance and priority of the task, and the variation value coefficient is obtained by weighted average.

[0048] By inputting the mutation value coefficient of each candidate solution into the robot control mutation function, combining the mutation value coefficient, generating mutation constraint rules, and establishing the qth mutation constraint rule. The mutation constraint rule determines how many times each candidate solution is mutated, that is, the number of mutations of each candidate solution corresponding to each robot control candidate solution. For example, solutions with higher mutation value coefficients will have more mutation operations. The mutation constraint rule ensures that the number and amplitude of mutation operations are reasonable, avoiding performance instability caused by excessive or insufficient mutations.

[0049] According to the generated q-th mutation constraint rule, mutate the control candidate domain of the q-th node robot, that is, perform a mutation operation on the control scheme in the q-th node task. By adjusting control parameters (such as action amplitude, timing, speed, etc.), a new control strategy is generated, forming the control mutation domain of the q-th node robot. The control mutation domain of the q-th node robot refers to the set of new control schemes generated through mutation operations in the q-th node task. For example, if the number of mutations is determined to be 5 according to the mutation constraint rule, five new control schemes will be generated and these five new candidate schemes will be added to the mutation domain.

[0050] Add the control mutation domain of the q-th node robot to the Q-dimensional mutation space of robot control. For other node tasks of the Q node tasks, perform the above steps to obtain the control mutation domains of the Q robots corresponding to the Q node tasks, which together constitute the Q-dimensional mutation space of robot control. The Q-dimensional mutation space of robot control refers to the high-dimensional space composed of all candidate control schemes generated through mutation in the Q node tasks. Each dimension represents a node task, and the control scheme mutates along these dimensions. By evaluating and constraining the mutation value of the candidate control scheme, more potential solutions are explored, increasing the possibility of finding a better solution, thereby comprehensively improving the collaboration effect and task completion accuracy.

[0051] Furthermore, this application also includes the following steps: The robot control mutation function is: ; where SDN represents the number of mutations of the candidate scheme, FLOOR means rounding down, SDK represents the predetermined number of mutations, SXC represents the mutation value coefficient of the candidate scheme, and SXO represents the predetermined mutation value coefficient of the candidate scheme.

[0052] Specifically, the robot control mutation function adjusts the predetermined number of mutations SDK according to the comparison result between the mutation value coefficient SXC of the candidate scheme and the predetermined mutation value coefficient SXO of the candidate scheme, so as to determine the final number of mutations SDN of the candidate scheme. SDN represents the number of mutations of the candidate scheme, indicating the number of mutations of the candidate scheme during the optimization process of the robot control strategy; FLOOR means rounding down, indicating rounding a value down to the nearest integer; SDK represents the predetermined number of mutations, that is, the number of mutations allowed in the initial setting; SXC represents the mutation value coefficient of the candidate scheme, indicating the degree of mutation of each candidate control decision; SXO represents the predetermined mutation value coefficient of the candidate scheme, the preset standard mutation value coefficient, used to compare with the mutation value coefficient of the candidate scheme.

[0053] When SXC < SXO, it indicates that the mutation value of the candidate solution is lower than expected. A decreasing function is used to adjust the number of mutations, aiming to control the number of mutations so that it is not too drastic and maintain a certain stability. The number of mutations is rounded down by the FLOOR function to ensure that the number of mutations is an integer. When SXC = SXO, the mutation value of the candidate solution exactly meets the expectation, and the number of mutations remains the predetermined number of mutations SDK, that is, no additional mutation adjustment is performed, and the mutations are carried out according to the preset number of mutation times. When SXC > SXO, the mutation value of the candidate solution is higher than expected, and the number of mutations increases, meaning that the mutation value of the current candidate solution is relatively large, and a greater mutation amplitude is required, which helps to accelerate the exploration of more control strategies to adapt to possible new task requirements or environmental changes.

[0054] Exemplarily, assume that the predetermined number of mutations is 10 times, the predetermined mutation value coefficient of the candidate solution is 5, and the mutation value coefficient of the current candidate solution is 3. At this time, SXC < SXO, so the first formula is used for calculation. By dynamically adjusting the number of mutations, the problems of excessive mutation or insufficient mutation are avoided, ensuring that the optimization process can effectively explore the optimization space and find a better solution.

[0055] S500: Execute the multi-limb collaborative control of the bionic robot according to the robot control optimization block.

[0056] Specifically, the robot control optimization block contains the optimization results of the task execution control schemes of multiple nodes and stores the key control information of task execution (such as the selection of control schemes, task completion accuracy, collaboration efficiency, etc.). By executing the optimization results in the robot control optimization block, the best control scheme for each node task is provided for the bionic robot. According to the optimization results in the robot control optimization block, a suitable control scheme is selected for each node task, and the multi-limb robot needs to collaborate according to the control scheme.

[0057] According to the best control scheme of each task result, determine the specific parameters of each scheme, including limb movements, timings, force control, etc. Convert the best control scheme into an instruction format that the robot can understand and send it to the limb control modules of the bionic robot to ensure that the instructions are distributed to the corresponding limbs in the correct order and time nodes. Each limb starts to execute actions according to the received instructions and monitors the movement state of the limb in real time to ensure the accuracy and coordination of the actions. During the task execution process, if real-time feedback or external environmental changes (such as changes in object weight, narrow space, etc.) affect the task progress, the robot will adjust the actions of each limb with reference to the control scheme in the optimization block.

[0058] After the actions of all task nodes are executed, the completion status of the entire task is evaluated, including task completion accuracy, limb collaboration efficiency, and limb collaboration reliability. Multi-limb collaboration control refers to the coordination and cooperation of multiple limbs (such as arms, legs, etc.) in a multi-limb robot system when performing complex tasks. Each limb adjusts its actions according to the environment and task requirements to ensure that the bionic robot can complete the task efficiently and accurately. By adopting the optimized control scheme, the bionic robot can precisely control the actions of each limb according to the task requirements and environmental conditions, ensuring the high-precision completion of the task. The collaborative control of multiple limbs can reduce ineffective or repetitive actions and improve the overall task execution efficiency of the robot. By using the optimal collaboration scheme, the robot can complete the task in a shorter time and avoid redundant actions.

[0059] In summary, the multi-limb collaboration control method for a bionic structure provided by this application has the following technical effects: By sequentially sorting out the tasks to be executed by the bionic robot and establishing a task chain to be executed, where the bionic robot includes N limbs, and the task chain to be executed includes Q node tasks, both N and Q are positive integers greater than 1; making collaborative control decisions for the N limbs according to the task chain to be executed and establishing a Q-dimensional robot control space; testing and optimizing the Q-dimensional robot control space according to the robot control test channel and establishing a Q-dimensional candidate space for robot control; introducing a robot control mutation function and a collaborative fitness analysis model to perform mutation expansion optimization on the Q-dimensional candidate space for robot control to obtain a robot control optimization block; performing multi-limb collaboration control of the bionic robot according to the robot control optimization block. That is to say, by sequentially sorting out the tasks to be executed and establishing a task chain, making collaborative control decisions for each limb according to the task chain, and selecting the optimal control strategy through the robot control test channel, the robot control mutation function, and the collaborative fitness analysis model for multi-limb collaboration control of the bionic robot, optimizing the collaborative actions between each limb, enhancing the adaptability of the multi-limb collaboration control strategy in a complex environment, and improving the task execution efficiency of the bionic robot.

[0060] Embodiment 2, based on the same inventive concept as the multi-limb collaboration control method for a bionic structure in the foregoing Embodiment 1, this application also provides a multi-limb collaboration control system for a bionic structure. Please refer to the appendix Figure 2 , the multi-limb collaboration control system for a bionic structure includes: The task timing sorting module 11 is used to sort the tasks to be executed by the bionic robot in terms of timing and establish a task chain to be executed. Among them, the bionic robot includes N limbs, and the task chain to be executed includes Q node tasks. Both N and Q are positive integers greater than 1; the collaborative control decision-making module 12 is used to make collaborative control decisions on the N limbs according to the task chain to be executed and establish a Q-dimensional robot control space; the inspection and optimization module 13 is used to inspect and optimize the Q-dimensional robot control space according to the robot control inspection channel and establish a Q-dimensional candidate space for robot control; the mutation expansion optimization module 14 is used to introduce a robot control mutation function and a collaborative fitness analysis model to perform mutation expansion optimization on the Q-dimensional candidate space for robot control and obtain an optimized block for robot control; the collaborative control execution module 15 is used to execute the multi-limb collaborative control of the bionic robot according to the optimized block for robot control.

[0061] Further, the collaborative control decision-making module 12 in the multi-limb collaborative control system of the bionic structure is further used for: According to the task chain to be executed, extract the q-th node task, where 1 ≤ q ≤ Q; collect the scene information of the q-th node task to obtain the q-th task scene; use the q-th task scene as the collaborative control scene constraint and the q-th node task as the collaborative control target to make control decisions on the N limbs and establish a control domain for the q-th node robot; add the control domain for the q-th node robot to the Q-dimensional robot control space.

[0062] Further, the inspection and optimization module 13 in the multi-limb collaborative control system of the bionic structure is further used for: Traverse the control domain for the q-th node robot, extract the first robot control decision; according to the first robot control decision, simulate the control of the bionic robot to obtain a first simulated collaboration data set; input the first simulated collaboration data set into the robot control inspection channel to obtain a first robot control inspection result; when the first robot control inspection result is qualified, set the first robot control decision as the first candidate solution for robot control and add the first candidate solution for robot control to the candidate domain for the q-th node robot; when the first robot control inspection result is unqualified, eliminate the first robot control decision; continue to inspect and optimize the control domain for the q-th node robot according to the robot control inspection channel, construct the candidate domain for the q-th node robot, and add the candidate domain for the q-th node robot to the Q-dimensional candidate space for robot control.

[0063] Further, the inspection and optimization module 13 in the multi-limb collaborative control system of the bionic structure is further used for: The robot control inspection channel includes a robot control evaluation model and a control inspection model; input the first simulated cooperation data set into the robot control evaluation model to obtain a first robot control evaluation result. Among them, the robot control evaluation model includes multi-dimensional robot control evaluation indicators, and the multi-dimensional robot control evaluation indicators include task completion accuracy, limb cooperation efficiency, and limb cooperation reliability; input the first robot control evaluation result into the control inspection model to output the first robot control inspection result. Among them, the control inspection model includes a control inspection operator. The control inspection operator includes that if the first robot control evaluation result meets the robot control inspection constraints, the first robot control inspection result is qualified; if the first robot control evaluation result does not meet the robot control inspection constraints, the first robot control inspection result is unqualified. The robot control inspection constraints include a predetermined task completion accuracy, a predetermined limb cooperation efficiency, and a predetermined limb cooperation reliability.

[0064] Further, the mutation expansion optimization module 14 in the multi-limb cooperation control system with a bionic structure is further configured to: Mutate the robot control Q-dimensional candidate space according to the robot control mutation function to establish a robot control Q-dimensional mutation space; perform inspection and optimization on the robot control Q-dimensional mutation space according to the robot control inspection channel to obtain a robot control Q-dimensional mutation optimization space; expand the robot control Q-dimensional candidate space according to the robot control Q-dimensional mutation optimization space to obtain a robot control Q-dimensional optimization space; perform cooperation fitness maximization optimization on the robot control Q-dimensional optimization space according to the cooperation fitness analysis model to obtain Q node robot control optimization results. Among them, the cooperation fitness analysis model includes a task completion accuracy weight, a limb cooperation efficiency weight, and a limb cooperation reliability weight; store the Q node robot control optimization results both on-chain and off-chain to generate the robot control optimization block.

[0065] Further, the mutation expansion optimization module 14 in the multi-limb cooperation control system with a bionic structure is further configured to: Connect to the robot control inspection channel, retrieve each robot control evaluation result corresponding to each robot control candidate solution within the robot control candidate domain of the q-th node; evaluate the mutation value of each robot control candidate solution based on the respective robot control evaluation results to obtain the mutation value coefficients of each candidate solution; input the mutation value coefficients of each candidate solution into the robot control mutation function to establish the q-th mutation constraint rule, where the q-th mutation constraint rule includes the mutation quantity of each candidate solution corresponding to each robot control candidate solution; mutate the robot control candidate domain of the q-th node according to the q-th mutation constraint rule to generate the robot control mutation domain of the q-th node, and add the robot control mutation domain of the q-th node to the robot control Q-dimensional mutation space.

[0066] Further, the mutation expansion optimization module 14 in the multi-limb collaborative control system with a bionic structure is further configured to: The robot control mutation function is: ; where SDN represents the mutation quantity of the candidate solution, FLOOR means rounding down, SDK represents the predetermined mutation quantity, SXC represents the mutation value coefficient of the candidate solution, and SXO represents the predetermined mutation value coefficient of the candidate solution.

[0067] In this specification, each embodiment is described in a progressive manner, and the key point of each embodiment is to illustrate the differences from other embodiments. The Figure 1 A bionic structure multi-limb collaborative control method and specific example in the first embodiment are equally applicable to the bionic structure multi-limb collaborative control system in this embodiment. Through the detailed description of the bionic structure multi-limb collaborative control method above, those skilled in the art can clearly understand the bionic structure multi-limb collaborative control system in this embodiment. Therefore, for the sake of brevity of the specification, it will not be elaborated here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0068] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0069] Obviously, those skilled in the art can make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of this application and its equivalent technologies, this application also intends to include these changes and modifications.

Claims

1. A multi-limb collaborative control method of a bionic structure, characterized in that: include: The tasks to be performed by the bionic robot are sequentially sorted to establish a task chain to be performed, wherein the bionic robot includes N limbs, the task chain to be performed includes Q node tasks, and N and Q are both positive integers greater than 1; Make collaborative control decisions on the N limbs according to the task chain to be executed, and establish a Q-dimensional robot control space; According to the robot control inspection channel, the Q-dimensional robot control space is inspected and optimized to establish a robot control Q-dimensional candidate space; Introducing a robot control mutation function and a collaborative fitness analytical model to perform mutation, expansion and optimization on the robot control Q-dimensional candidate space to obtain a robot control optimization block; The multi-limb collaborative control of the bionic robot is performed according to the robot control optimization block.

2. A multi-limb collaborative control method of a bionic structure as claimed in claim 1, characterized in that: According to the task chain to be executed, collaborative control decisions are made on the N limbs to establish a Q-dimensional robot control space, including: According to the task chain to be executed, extract the qth node task, where 1≤q≤Q; Collecting scene information of the qth node task to obtain the qth task scene; Taking the qth task scenario as the collaborative control scenario constraint and the qth node task as the collaborative control target, making control decisions on the N limbs and establishing a qth node robot control domain; Add the q-th node robot control domain to the Q-dimensional robot control space.

3. The multi-limb cooperative control method of a bionic structure as claimed in claim 1, characterized in that: The Q-dimensional robot control space is tested and optimized according to the robot control test channel to establish a robot control Q-dimensional candidate space, including: Traverse the robot control domain of the qth node and extract the first robot control decision; According to the first robot control decision, simulate and control the bionic robot to obtain a first simulated collaboration data set; Inputting the first simulated collaborative data set into the robot control inspection channel to obtain a first robot control inspection result; When the first robot control inspection result is qualified, setting the first robot control decision as a first robot control candidate solution, and adding the first robot control candidate solution to the qth node robot control candidate domain; When the first robot control inspection result is unqualified, eliminating the first robot control decision; The q-th node robot control domain is continuously tested and optimized according to the robot control test channel, the q-th node robot control candidate domain is constructed, and the q-th node robot control candidate domain is added to the robot control Q-dimensional candidate space.

4. A multi-limb collaborative control method of a bionic structure as claimed in claim 3, characterized in that: Inputting the first simulated collaborative data set into the robot control inspection channel to obtain a first robot control inspection result includes: The robot control inspection channel includes a robot control evaluation model and a control inspection model; Inputting the first simulated collaboration data set into the robot control evaluation model to obtain a first robot control evaluation result, wherein the robot control evaluation model includes a robot control multidimensional evaluation index, and the robot control multidimensional evaluation index includes task completion accuracy, limb collaboration efficiency, and limb collaboration reliability; The first robot control evaluation result is input into the control verification model, and the first robot control verification result is output, wherein the control verification model includes a control verification operator, and the control verification operator includes: if the first robot control evaluation result satisfies the robot control verification constraint, the first robot control verification result is qualified; if the first robot control evaluation result does not satisfy the robot control verification constraint, the first robot control verification result is unqualified, and the robot control verification constraint includes a predetermined task completion accuracy, a predetermined limb collaboration efficiency, and a predetermined limb collaboration reliability.

5. The multi-limb cooperative control method of a bionic structure as claimed in claim 1, characterized in that: The robot control mutation function and the collaborative fitness analytical model are introduced to perform mutation expansion optimization on the robot control Q-dimensional candidate space to obtain the robot control optimization block, including: mutating the robot control Q-dimensional candidate space according to the robot control mutation function to establish a robot control Q-dimensional mutation space; According to the robot control inspection channel, the robot control Q-dimensional variation space is inspected and optimized to obtain the robot control Q-dimensional variation optimization space; Expanding the robot control Q-dimensional candidate space according to the robot control Q-dimensional variation optimization space to obtain the robot control Q-dimensional optimization space; According to the collaborative fitness analytical model, the collaborative fitness maximization optimization is performed on the Q-dimensional optimization space of the robot control to obtain the optimization results of the Q-node robot control, wherein the collaborative fitness analytical model includes a task completion accuracy weight, a limb collaboration efficiency weight, and a limb collaboration reliability weight; The control optimization results of the Q node robots are stored on and off the chain to generate the robot control optimization block.

6. A multi-limb cooperative control method of a bionic structure as claimed in claim 5, characterized in that: The robot control Q-dimensional candidate space is mutated according to the robot control mutation function to establish a robot control Q-dimensional mutation space, including: Connecting to the robot control inspection channel, retrieving each robot control evaluation result corresponding to each robot control candidate solution in the q-th node robot control candidate domain; Performing variation value evaluation on each robot control candidate solution according to each robot control evaluation result to obtain a variation value coefficient of each candidate solution; Inputting the variation value coefficient of each candidate solution into the robot control variation function to establish the qth variation constraint rule, wherein the qth variation constraint rule includes the variation quantity of each candidate solution corresponding to each robot control candidate solution; The q-th node robot control candidate domain is mutated according to the q-th mutation constraint rule to generate a q-th node robot control mutation domain, and the q-th node robot control mutation domain is added to the robot control Q-dimensional mutation space.

7. The multi-limb cooperative control method of a bionic structure as claimed in claim 1, characterized in that: The robot control variation function is: ； Among them, SDN represents the number of candidate solution variations, FLOOR refers to rounding down, SDK represents the predetermined number of variations, SXC represents the candidate solution variation value coefficient, and SXO represents the predetermined candidate solution variation value coefficient.

8. A bionic structure multi-limb cooperative control system, characterized in that: Steps for implementing a multi-limb cooperative control method of a bionic structure as claimed in any one of claims 1 to 7, wherein the multi-limb cooperative control system of a bionic structure comprises: A task timing combing module is used to perform timing combing on the tasks to be executed of the bionic robot and establish a task chain to be executed, wherein the bionic robot includes N limbs, the task chain to be executed includes Q node tasks, and N and Q are both positive integers greater than 1; A collaborative control decision module, used to make collaborative control decisions on the N limbs according to the task chain to be executed, and establish a Q-dimensional robot control space; A test and optimization module, used for testing and optimizing the Q-dimensional robot control space according to the robot control test channel, and establishing a robot control Q-dimensional candidate space; A mutation expansion optimization module is used to introduce a robot control mutation function and a collaborative fitness analytical model to perform mutation expansion optimization on the robot control Q-dimensional candidate space to obtain a robot control optimization block; A collaborative control execution module is used to execute multi-limb collaborative control of the bionic robot according to the robot control optimization block.

Images