基于遗传算法的机器人运动学参数误差识别

Robotic Kinematics Parameters Error Identification Based on Genetic Algorithm

针对机器人标定过程中误差建模与参数识别方法的繁琐问题,提出通过考虑参数误差的运动学模型直接建立基于相对距离误差的参数误差方程组,并采用混合遗传算法求解方程组的新方法,使得参数误差识别简单易行。算例通过数值仿真,分析了采用不同的测量数据(完整的距离信息或非完整的距离信息)进行参数误差识别的可行性,结果表明,只要获得足够多的测量数据,就可以识别机器人运动学参数误差,这为机器人标定时采用一些非完整的距离测量方法提供了理论支持。

Aimed at the tedious error modeling and parameters identification in the process of robot calibration, establishing parameter error equations directly based on relative distance error using the kinematics model considering the parameters error was proposed. The hybrid genetic algorithm was adopted to solve the equations. As the given calculation example, the feasibility of identification parameters error using different measured data （complete distance info or incomplete distance info） was analyzed by numerical simulation. The results show that the robotic kinematics parameters error can be obtained if the data were measured enough. This result provides a theoretic support for using of incomplete distance measuring method in robot calibration process.