基于遗传算法的机械臂逆运动学问题解决方案

Solution based on genetic algorithm for inverse problem of manipulator kinematics

提出应用遗传算法求解机械臂的逆运动学问题,将种群定义于机械臂的关节角轨迹层面,利用连续性函数实现算法的初始化算子,交叉算子和变异算子。算法仅使用表现型数据表示方式,克服了传统遗传算法在数据的基因型和表现型之间频繁地进行编码和解码操作。通过和传统遗传算法进行对比分析,验证了所提出的方法能够避免传统遗传算法求解逆运动学问题时存在的多重切换点现象,能够获得更平滑的关节角轨迹,缩短了算法的收敛时间,生成的笛卡尔轨迹具有更高的精度。

In this paper, genetic algorithm is applied to the inverse problem of manipulator kinematics and the population is defined on the manipulator joints path level. The initialization operator, crossover operator and mutation operator of the algorithm are implemented by continuous function. The algorithm uses only the phenotype data presentation, which overcomes the conventional genetic algorithm frequently carrying out coding and decoding operations in the data between the genotype and phenotype. With the conventional genetic algorithm carried on the contrast analysis, it is observed that the proposed method can avoid the disadvantage of the conventional genetic algorithm having multiple switching points for solving the inverse kinematics problem, obtain a smoother joint angle trajectory, shorten the convergence time of the algorithm, and the Cartesian trajectory generated has higher accuracy.