High Performance Control of Robot Manipulator without Inverse Dynamics

点击翻译按钮获取中文摘要

Inverse dynamics has been a necessary technique in high pcrformance robot control. In this research, we proposed the new paradigm of robot control without using inverse dynamics. The proposed method is based on the robust servo controller implemented in each joint. In actuality, we showed that this method is effective enough even in the control of DD(Direct Drive)manipulator, where the interference force among joints are very strong(1) Robust Joint Servo Controller13EA03 : We proposed the Two-Degree-Of-Freedom controller to improve the disturbance rejection and robustness to parameter variation. We clarified the relation among various controllers. We investigated the stability problcm of robust servo system and proved that the proposed method can be realized with enough stability using very simple control hardware. Further, the instantaneous speed observer which realizes the high robustness even when the encoder pulse resolution is small.(2) Unified Method to Realize Various Robot Cont … More rolBased on the proposed unified approach, we realized almost all types of robot control, ie., impedance, compliant motion, force and hybrid controls. We showed their effectiveness by experiments.(3) Robust Force Control We realized the novel method to grasp various objects by the robust controller with fixed coefficients without using any adaptive or learning techniques.(4) Collision ControlWe developed the novel method to reduce the impact force caused by collision when trajectory control is changed to force control. We used a force sensor or a simple disturbance observer to detect the collision. We also developed the high speed algorithm to perform the collision-free path planning.(5) Control of Flexible JointBased on the 2-inertia model to represent the controlled object, we proposed various methods to suppress the vibration : state observer, H* controller, etc. In particular, we proposed the resonance ratio control using the disturbancc observer and showcd its effectiveness by simulation and experiment. Less