计算机工程与应用 ›› 2017, Vol. 53 ›› Issue (8): 126-131.DOI: 10.3778/j.issn.1002-8331.1510-0175

• 模式识别与人工智能 • 上一篇    下一篇

机器人逆运动学差分自适应混沌粒子群求解

谢  宏,向启均,陈海滨,张小刚,杨  鹏,张爱林,李云峰   

  1. 湖南大学 电气与信息工程学院 自动化系,长沙 410082
  • 出版日期:2017-04-15 发布日期:2017-04-28

Inverse kinematics solution algorithm of robot based on differential algorithm and adaptive chaotic particle swarm optimization

XIE Hong, XIANG Qijun, CHEN Haibin, ZHANG Xiaogang, YANG Peng, ZHANG Ailin, LI Yunfeng   

  1. Department of Automation, College of Electrical and Information Engineering, Hunan University, Changsha 410082, China
  • Online:2017-04-15 Published:2017-04-28

摘要: 采用D-H法通过连杆坐标系变换矩阵建立机械臂运动控制模型,该模型呈现非常严重的非线性特性,传统方法难以求解。由于动态差分算法具有很强的全局搜索能力,而粒子群算法具有精确的局部搜索能力的特点,融合改进的动态差分算法和粒子群算法,并引入混沌映射初始种群和粒子群学习因子与惯性权重的自适应算法,提出多子群分层差分自适应混沌粒子群算法。该算法采用的多子群分层结构能提升个体共享群体信息的能力,底层利用动态差分算法进行全局搜索,顶层精英群利用改进的粒子群算法进行局部搜索。仿真试验和实际应用表明该算法在稳定性、搜索成功率以及收敛精度有显著提高,能有效解决机器人逆运动学模型的求解。

关键词: 逆运动学, 混沌粒子群, 差分算法, 多子群分层

Abstract: The mechanical arm control model established by using the D-H method of transformation matrix is seriously nonlinear, and it is difficult to solve through the traditional methods. The dynamic differential algorithm has the great global searching ability and the particle swarm optimization algorithm has the significant local searching accuracy. The paper mixes the improved dynamic differential algorithm and the particle swarm optimization algorithm, and adoptes the chaotic algorithm to initialize individuals’ values and the adaptive algorithm to modify the study facts and weight facts of particle swarm optimization. It presents multi-sub group hierarchical hybrid algorithm of differential algorithm and adaptive chaotic particle swarm optimization. The multi-subgroup hierarchical hybrid structure adopted by the paper can improves the ability of sharing the information among the individuals. The bottom structure uses the dynamic differential algorithm for the global searching, and the top elite population uses the improved adaptive chaotic particle swarm optimization for the local searching. The experimental simulation and practice result prove that the stability, the successful rate of searching and the convergence precision have been greatly improved. The discussed algorithm can solve the problem of the inverse kinematics model efficiently.

Key words: inverse kinematics, chaotic particle swarm optimization, differential algorithm, multi-subgroup hierarchical