在人机协作中基于动量观测和优化算法的全机械臂单点接触信息实时估计

Real-time Estimation of Single Contact Information of the Whole Robotic Arm Based on the Momentum Observation and Optimization Algorithm in Human-Robot Cooperation

针对协作型机器人,提出了一种依据机器人控制与运动状态信息,基于动量观测与优化算法相结合的全机械臂单点接触信息实时估计方法.该方法将接触位置估计问题转化为优化算法有界搜索问题,搜索范围由动量观测方法确定的接触臂长度确定,两种方法的结合保证了估计的精度和实时性.基于UR机械臂实验平台的仿真与实验结果表明,本文方法能够实时、准确地估计除第1关节以外首次发生接触的机械臂任意位置处的力和位置信息.本研究可以在不借助传感器的情况下实现机械臂对外力的感知.

For collaborative robots, a real-time method is proposed estimate the single contact information of the whole robotic arm according to the information of control system and motion state based on the momentum observation and optimization algorithm. The proposed method transforms the estimation problem of contact position into bounded searching problem of the optimization algorithm. The searching range is determined by the length of contact robotic arm, which is obtained through momentum observation. The combination of the above two methods can guarantee the precision and speed of estimation. The results of simulation and experiment based on UR robotic arm test platform show that the proposed method can accurately estimate single contact information in real time which occurs for the first time at anywhere of the robotic arm except the first joint. Based on this research, the robotic arm can perceive external forces without sensors.