以比例閥控制雙氣壓人造肌肉致動器應用於下肢二自由度機械之研究

Abstract

隨著全球進入高齡化社會，外骨骼在機器人領域越來越受歡迎，因為它們不僅可以對病人受損的四肢提供協助也能輔助難以自己移動的老年人，同時也能增加健全人的力量。在各種的致動器之中氣壓人造肌肉致動器(pneumatic artificial muscles actuator)對於外骨骼設備而言具有很大的優勢，由於其固有的可彎曲性，這保證了操作者和設備之間的安全。此外該高出力-重量比和重量輕也是外骨骼設備所需要的理想特性。但因其高度非線性的缺點，使得高控制精度不易。 本論文旨在以氣壓人造肌肉致動器應用於雙自由度下肢機器人系統，作為未來朝向下肢外骨骼輔具發展之起始。本文之下肢機器人系統分為大腿和小腿兩軸，各軸分別以比例閥控之雙氣壓人造肌肉致動器驅動，進行類似於人類下肢之雙自由度下肢機器人系統設計及實驗原型系統建立，考慮行動性及重量，本文使用比例閥來取代壓力閥。由於氣動人工肌肉致動器是高度非線性的致動器，所以不易建立準確的數學模型，因此採用不需系統模式之非線性類神經自調式PID控制器。此外，雙自由度機械人也須發展正逆向運動學分析。而角度感測器為類比輸出使的訊號產生嚴重的干擾，所以多加設計了干擾觀測器來擷取角度訊號可以減少雜訊的干擾。實驗先分別實現單軸角度軌跡控制後，再進一步整合雙軸運動及軌跡規劃，進行雙自由度下肢機器人系統在不同軌跡下的運動控制。最後，實驗結果顯示基於類神經網路的自調式PID控制器可有效控制比例閥運用在雙自由度下肢機器人系統，由於此下肢機器人系統未來將朝向外骨骼發展，所以系統的動作流暢性會比精準度來的重要，因此，本研究的實驗控制誤差可保持在可接受的範圍內。

Rehabilitation robots and exoskeletons have increasingly become popular in the field of robotics, since they can not only provide a support for patients with impaired limbs or the elders with difficulty of doing activities for daily living by their own, but also augment the power of able-bodied people. Of all the actuators, pneumatic artificial muscles (PAMs) may be the most promising one due to their inherent compliance, which guarantees safe interactions between the operator and the device. In addition, high power to weight ratio and lightness are also ideal features for the applications of human-friendly devices. However, the nonlinearity is the drawback that is required to mitigate for accurate control. The purpose of this study is to develop a dual-PAMs driving 2-DOF robotic system, following with the research of [1] for our future objective of the lower limb rehabilitation robot. The system structure is similar to a human lower limb. The test rig of the dual-PAMs driving 2-DOF robotic system is composed of upper leg, lower leg, and each leg is equipped with a proportional-valve controlled dual-PAMs to reduce the system weight. Since the PAMs is a highly non-linear actuator, it is hard to control the system and derive mathematical model precisely. Therefore, the system is controlled by the modified model-free self-tuning PID controller based on neural network to compensate the nonlinearity and improve the tracking performance. For the 2-DOF motion of lower limb, kinematics and inverse kinematics are derived. Finally, the experimental results indicate that 2-DOF tracking motion control of lower limb of the dual-PAMs driving 2-DOF robotic system can be achieved by the self-tuning PID controller with acceptable control error.