基于量子粒子群优化算法的机器人运动学标定方法  

Kinematic Calibration Method of Robots Based on Quantum-behaved Particle Swarm Optimization

作　　者：房立金[1] 党鹏飞[1] 

机构地区：[1]东北大学机械工程与自动化学院,沈阳110819

出　　处：《机械工程学报》2016年第7期23-30,共8页Journal of Mechanical Engineering

基　　金：国家自然科学基金(51575092);辽宁重大装备制造协同创新中心资助项目

摘　　要：基于量子粒子群优化算法,提出一种同样适用于串联机器人和并联机器人的运动学标定方法。利用闭环矢量链方法和Denavit-Hartenberg矩阵法,分别建立并联机器人和串联机器人的运动学误差模型,将运动学误差模型内的几何误差源作为相应的机构参数修正量。由于机器人运动学误差模型表现有较强的非线性,因此确定模型内的机构参数修正量为优化变量,将机器人运动学参数标定问题转化为非线性系统的优化问题。采用量子粒子群优化算法对优化问题进行求解,利用优化获得的参数修正量更新运动学模型,以达到提高机器人运动精度的目的。以五轴并联机床的平面约束机构为研究对象,通过试验验证该标定方法的标定效果,并与模糊插值标定方法进行比较分析,结果表明在较大的工作空间内基于量子粒子群优化的运动学标定方法更为有效。Based on quantum-behaved particle swarm optimization（QPSO）, a kinematic calibration method is proposed, which is applicable to both serial robots and parallel robots. Based on closed-loop vector chain method and Denavit-Hartenberg method, The kinematic error models of these two robots are built and the robots＇ mechanism parameters are determined as optimal variables. Due to the nonlinearity of kinematic error models, the kinematic parameters calibration problem is transformed to a nonlinear optimization problem. The QPSO algorithm is applied to search for the global optimum solution of the optimization problem, update the kinematic models with calculation results to increase robots＇ accuracy. The effectiveness of the proposed calibration approach is demonstrated through some experiments based on the constraint mechanism of 5-DOF parallel machine tool. Compared with fuzzy interpolation method, this kinematic calibration method based on QPSO is more effective in a biggish workspace.

关 键 词：机器人 位姿误差 运动学标定 量子粒子群优化 

分 类 号：TG502[金属学及工艺—金属切削加工及机床]