基于多种群遗传算法的一般机器人逆运动学求解

发布时间：2018-05-03 11:22

本文选题：一般机器人 + 运动学逆解　；参考：《机械工程学报》2017年03期

【摘要】：几何结构不满足Pieper准则的机器人被称为一般机器人,其逆运动学运算不能采用封闭解法,而采用数值解法又需要庞大的计算量,且存在奇异位置无法求解的问题。为此,将多种群遗传算法应用于运动学逆解运算,提出一种适用于一般机器人的高精度并行求逆算法。为避免机器人位姿收敛精度不同,该算法将目标函数分解为位置和姿态函数,同时引入适应度函数权值系数来平衡两函数收敛速度;为避免局部收敛,该算法采用多点交叉和均匀交叉相结合的交叉算子,并逐步增大均匀交叉概率来抑制短子串偏差,使搜索趋于稳健;为提高收敛速度,该算法采用动态变异率的变异算子,以及种群替代和个体替代相结合的移民算子来克服全局收敛的盲目性。以封闭解法和数值解法无法求逆的6R一般机械臂为对象,开展与单种群遗传算法的对比试验,结果表明:该算法可在避免局部收敛的基础上保证算法稳定性,且能够大幅提升收敛精度和速度。
[Abstract]:A robot whose geometric structure does not meet the Pieper criterion is called a general robot. The inverse kinematics can not be solved by a closed method, but the numerical method requires a large amount of computation, and there is a problem that the singular position cannot be solved. For this reason, the multi-population genetic algorithm is applied to kinematics inverse solution operation, and a high precision parallel inverse algorithm is proposed for general robot. In order to avoid the different convergence accuracy of robot pose, the objective function is decomposed into position and attitude functions, and the fitness function weight coefficient is introduced to balance the convergence rate of the two functions. The algorithm adopts the crossover operator which combines multi-point crossover and uniform crossover, and gradually increases the uniform crossover probability to suppress the short substring deviation and make the search tend to be robust, in order to improve the convergence speed, the algorithm adopts the mutation operator with dynamic mutation rate. And the migration operator which combines population substitution and individual substitution to overcome the blindness of global convergence. Taking the general 6R manipulator which can not be inversed by the closed and numerical methods as the object, the comparison experiment with the single population genetic algorithm is carried out. The results show that the algorithm can guarantee the stability of the algorithm on the basis of avoiding the local convergence. And can greatly improve the convergence accuracy and speed.
【作者单位】：华中科技大学数字制造装备与技术国家重点实验室;
【基金】：国家自然科学基金(51535004,51323009) 中国博士后科学基金(2014M562013,2015T80789)资助项目
【分类号】：TP18;TP242

【参考文献】