Fluid Series Elastic Force Control for Robot Manipulation

Chunpeng Wang, Zikun Yu,Eric Schwarm, Xiao Huang

semanticscholar(2019)

引用 0|浏览0
暂无评分
摘要
Fluid-based actuation is an attractive option for soft, lightweight, and human-safe robots. Combining with force control, we would like to make this kind of system with the ability to provide the widest possible range of contact actions, from soft and gentle to stiff and strong, while maintaining contact stability with any environment. In this thesis, each motor-hydraulic-finger system of our fluid actuated gripper is modelled as a classical 2-DOF series-elastic actuation (SEA) system. The low-friction and low-hysteresis mechanical properties of rolling diaphragm fluid actuators offer the ability to apply force control based on internal pressure force feedback targeting to eliminate motor nonlinear friction. Using internal pressure force and motor state feedback, an expanded range of stability is achieved without sacrificing high closed-loop bandwidth and passivity. Several force control methods, including disturbance observer (DOB), natural admittance control (NAC), and integral force and motor state feedback, are discussed and shown to be equivalent. Based on the knowledge above, a DOB-based impedance controller for the 1-DOF motor system is introduced to extend endpoint impedance range. Endpoint low impedance can be further reduced, while the ability to render the largest possible impedance is maintained. Contact stability with environment is guaranteed by showing the motor system alone is passive. Results show that the motor plant can be regulated as either the pure physical motor inertia or the largest possible equivalent virtual spring for rendering either low impedance or high impedance. Applying an additional model-based feed-forward friction compensation, with passivity consideration for linear damping and stiffness compensation, endpoint impedance dynamic range at low i frequency region using our controller is 50dB with 17dB increase compared to the case without force control.
更多
查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要