模块化机器人构型多目标优化设计  

Multi-objective Optimization Design of Modular Robot Configurations

作　　者：周育华 甘文峰 廖昭洋 黄海滔 韦文斐 陈烽民 Zhou Yuhua;Gan Wenfeng;Liao Zhaoyang;Huang Haitao;Wei Wenfei;Chen Fengmin(School of Electromechanical Engineering,Guangdong University of Technology,Guangzhou 510006,China;ZWSOFT Co.,Ltd.,Guangzhou 510623,China;Institute of Intelligent Manufacturing,Guangdong Academy of Sciences,Guangzhou 510070,China;School of Mechanical and Automotive Engineering,South China University of Technology,Guangzhou 510640,China)

机构地区：[1]广东工业大学机电工程学院,广州510006 [2]广州中望龙腾软件股份有限公司,广州510627 [3]广东省科学院智能制造研究所,广州510070 [4]华南理工大学机械与汽车工程学院,广州510640

出　　处：《机电工程技术》2025年第4期142-145,160,共5页Mechanical & Electrical Engineering Technology

基　　金：广州市重点领域研发计划2021年度“新一代信息技术”重大科技专项(202103020004)。

摘　　要：为解决如何根据特定任务需求快速设计机器人的问题,提出一种性能驱动的模块化机器人构型多目标优化设计方法。以总质量、单位工作空间、全域可操作度3个性能指标为优化目标,构建模块化机器人构型的多目标优化模型。针对在构型优化问题中,连杆尺寸选择过少的问题,对优化变量采用整数编码方案,使连杆尺寸能在范围内连续取值,将离散尺寸与连续尺寸的构型优化实例进行对比,通过计算pareto前沿中所有点的欧氏距离均值衡量解集的分散度,结果表明尺寸取值连续实例的解集分散度较取值分散的实例减少了12.89%,验证连续尺寸方案的有效性。对pareto前沿中的构型进行分析,采用皮尔森相关系数法判断各优化目标之间的相关性,结果表明单位工作空间与全域可操作度的相关性最高,达到0.701,总质量与单位工作空间的相关性仅有0.007。在MATLAB上选择解集中的两个构型进行工作空间仿真分析,仿真结果表明,在总质量相同的情况下,构型A相较于构型B的单位长度工作空间减少了12.86%,同时全域可操作度增加了5.24%,该方法设计得到的解集中的构型均能满足任务需求。To address how to quickly design robots for specific tasks,a performance-driven multi-objective optimization method for modular robot configuration is proposed.A multi-objective optimization model for modular robot configurations has been constructed,with total mass,unit workspace,and global manipulability.Addressing the issue of limited link dimension choices in configuration optimization,an integer coding scheme for optimization variables has been adopted,allowing continuous link dimension selection within a specified range.By comparing optimization instances of discrete and continuous dimensions and measuring the dispersion of the solution sets through the average Euclidean distance among all points on the pareto front,results show a 12.89%reduction in dispersion for continuous dimension instances,confirming the effectiveness of the continuous dimension approach.Analysis of configurations on the pareto front using the Pearson correlation coefficient reveals the highest correlation between unit workspace and global manipulability at 0.701,while the correlation between total mass and unit workspace is only 0.007.Workspace simulation analysis of two configurations from the solution set in MATLAB shows that,with the same total mass,configuration A reduces the unit length workspace by 12.86%compared to configuration B but increases global manipulability by 5.24%.The configurations designed by this method in the solution set all meet the task requirements.

关 键 词：模块化机器人 遗传算法 构型优化 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]