基于可变遗忘因子递推最小二乘法的无传感器机器人末端接触力估计

Sensorless Robot End-Effector Contact Force Estimation Based on VFF-RLS

随着人机共融生产模式的推广,人与机器人需要协作完成工作任务,在人机协作的过程中需要估计机器人末端接触力。传统的机器人末端接触力估计主要是基于外部传感器来实现的,这不仅会使机器人本体成本增加,还会使机器人的控制系统变得更加复杂。针对这个问题,研究了无外部传感器的机器人末端接触力估计算法。首先设计数字低通滤波器对机器人动力学方程进行滤波,建立不显含加速度信号的机器人动力学模型,然后将机器人末端接触力看作时变参数,采用递推最小二乘法估计末端接触力,通过动态的改变遗忘因子使算法具有更好的响应特性,达到了较好的效果。最后通过MATLAB和ADAMS联合仿真验证了算法的有效性。

With the development of collaborative production, human and robot need to co-operate in work, so it is necessary to estimate the robot end-effector contact force. However, conventional contact force estimation methods are based on additional sensors, which are not only enhance cost but also increase the complexity of system. To solve this problem, we propose a contact force estimation method without any extra sensors. First, filtering the robot dynamics equation and establish the dynamics equation without acceleration, then we consider the contact force as time-varying parameters and estimate the contact force based on recursive least-squares algorithm, by changing the forgetting factor adaptively, the algorithm achieve good response characteristics. Finally, the results of simulation on MATLAB and ADAMS verify that the algorithm is valid.