一种改进的机器人路径优化算法  

An improved robot path optimization algorithm

作　　者：闫雪超 童东兵[1] 陈巧玉 YAN Xuechao;TONG Dongbing;CHEN Qiaoyu(College of Electronic and Electrical Engineering,Shanghai University of Engineering Science,Shanghai 201620,China;School of Statistics and Mathematics,Shanghai Lixin University of Accounting and Finance,Shanghai 201620,China)

机构地区：[1]上海工程技术大学电子电气工程学院,上海201620 [2]上海立信会计金融学院统计与数学学院,上海201620

出　　处：《传感器与微系统》2020年第2期109-112,共4页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(61673257,11501367);上海工程技术大学模式识别与智能系统学科建设项目(2018xk-B-09);上海工程技术大学研究生科研创新资助项目(17KY0209)

摘　　要：针对遗传算法(GA)的早熟收敛和随机变异导致不可行路径的问题,提出一种用于机器人路径规划的改进遗传算法。采用一种新的变异算子,根据总体适应度调整突变节点,保证突变产生可行路径,从而提高种群多样性,避免过早收敛。首先,采用栅格法建立机器人路径规划模型。其次,优化初始种群,得到不包含与障碍物相交的初始种群。最后,在适应度函数中引入惩罚函数系数,对不可行路径给予高惩罚,确保算法工作在可行路径。仿真结果表明:所提算法不仅优于传统遗传算法,还优于某些改进的遗传算法。并且可以在静态、动态环境下实现全局和局部路径规划,得到机器人最优或次优路径。Aiming at the problem that premature convergence and random mutation of genetic algorithm(GA)lead to infeasible path,an improved genetic algorithm for robot path planning is proposed.The algorithm uses a new mutation operator,which adjusts the mutation nodes according to the overall fitness to ensure that the mutation produces a feasible path,so as to increase population diversity and avoid premature convergence.Firstly,the robot path planning model is established by the grid method.Next,optimize the initial population structure so that the initial population does not intersect the obstacle.Finally,the penalty function coefficient is introduced in the fitness function,and high penalty is given to the infeasible path to ensure the algorithm works on the feasible path.The simulation results show that the proposed algorithm not only outperforms the traditional genetic algorithm but also outperforms some improved genetic algorithms.Using this algorithm,global and local path planning can be implemented in static and dynamic environments,and the optimal or suboptimal paths of the robot can be obtained.

关 键 词：路径规划 栅格法 变异算子 改进遗传算法 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]