摘要
提出了一种基于学习的动力学前馈方法 ,解决了欠驱动系统由于系统不完全可控造成的反馈控制失效、运动轨道难以实现的问题 .在此基础上提出了学习 -优化交替式最优运动轨道生成方法 ,该方法将轨道优化与学习控制相结合 ,所生成轨道可以在欠驱动机器人上实际操作 .对一台实际的欠驱动机器人实现了最优轨道生成 ,并进行了动作实验 。
A method of dynamics feedforward control based on learning was proposed. It solved the problem that the smooth feedback control is invalid for under actuated robots, and can drive an underactuated robot in its trajectory accurately. A learning optimizing alternated method was proposed to generate optimal trajectories for under actuated robots. It is realizable for a real system. The effectiveness of the method is confirmed by an experiment with a real under actuated robot.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2002年第10期1422-1425,1434,共5页
Journal of Shanghai Jiaotong University
