基于改进型遗传算法的虚拟人上肢运动链逆运动学求解方法

由于人体上肢运动链的高自由度,用传统的几何法、解析法、迭代法等求其逆解较为困难。遗传算法具有很好的寻优特性,但标准遗传算法在求解时容易陷入早熟收敛和后期搜索迟钝。为此,提出了一种改进型遗传算法(IGA)求解的方法。先构建人体上肢运动链的各关节单元,并用D-H方法建立其数学模型;然后仿人类种群现象实现遗传算法的种群多样化和种群初始化,设计具有自适应性能的交叉概率和变异概率算子,从而完成了对标准遗传算法的改进。通过对比仿真计算结果可得,改进后的遗传算法能以更大概率避免陷入早熟收敛和后期搜索迟钝,并以较少的遗传代数寻得高精度逆解。

Solution method for inverse kinematics of virtual human's upper limb kinematic chain based on improved genetic algorithm

An Improved Genetic Algorithm (IGA) was proposed for the inverse kinematics problem solution of upper limb kinematic chain which had high degree of freedom and was too complex to be solved by using geometric, algebraic, and iterative methods. First, the joint-units of upper limb kinematic chain and its mathematical modeling were constructed by using Denavit-Hartenberg (D-H) method, then population diversity and initialization were completed based on simulating human being population, and the adaptive operators for crossover and mutation were designed. The simulation results show that the IGA can search the high precise solutions and avoid prematurity convergence or inefficient searching in later stage with larger probability than standard genetic algorithm.