基於能量最佳化抬舉輔助機器人之軌跡規劃

Abstract

現今全球大部分已開發國家逐漸經邁入了老年化社會，再加上零抬舉照護政策的推廣，輔具機器人的開發已日漸受到重視。本研究的主要目標為探討並發展一台抬舉輔助機器人協助行動不便者由坐到站的最省能量的控制軌跡，主藉由沿著一條最省能量的軌跡行進可以讓整體運動過程節省能量的成本。本研究分別採用三自由度的輔具機器人動態模型以及四個自由度Dobot機械手臂進行最省能量軌跡規劃模擬分析與實測驗證。 本篇論文首先蒐集護理人員扶助行動不便者由坐到站的運動軌跡數據，並轉換為本研究協助由坐到站輔具機器人模型控制，並利用Newton-Euler理論為基礎推導出輔具機器人動力學方程式，接著再把推導完成的適度函式(fitness function)寫入MATLAB，然後分別利用GA演算法與PSO演算法找出其最省能量軌跡的最佳時間區間參數；在此同時對所要研究的協助由坐到站輔具機器人先利用SolidWorks進行幾何建模，然後再把建構好的模型匯入多體動力學模擬軟體RecurDyn進行動態模擬；期間分別採用GA演算法與PSO演算法解出的最省能量的軸運動軌跡輸入協助由坐到站輔具機器人在RecurDyn的動態模型以確認最省能量源軌跡的數學模型與參數，分析的結果與應用MATLAB計算出來的各軸馬達能量損耗結果加以比較均驗證本研究所漸粒的控制軌跡有最省能源的效益。實驗驗證方面，採用Dobot機械手臂進行不同運動軌跡驅動的動態能量損耗的實測分析驗證，實驗結果與MATLAB模擬及RecurDyn模擬的結果比較，均有一致的結論。 由模擬與實驗分析可知經過最佳化規劃後的動態軌跡的確有比一般任選的路徑達到了更省能量的結果，不管由採用實驗還是RecurDyn與MATLAB計算與模擬出來的結果均類似。本研究結果可以對之後要進行多自由度輔具機器人模型的研究或者進行移動物品的最省能量規劃研究驗證的參考。

Since that the majority of the developed countries in the world come to the social issue of aging population. Furthermore, no-lift policy in rehabilitation process is token attention and already legislated in some countries. Currently, various of the assistive technologies of rehabilitation robot in different applications have been developed and discussed. This study focuses on the development of sit-to-stand assistance robot considering energy-optimal trajectory planning for reducing the operation cost.

First of all, this study presents the dynamic equations of sit-to-stand assistance robot derived by using Newton-Euler method. The fitness function regarding the energy cost is derived in the form as the product of torque and angular velocity. Both of the genetic algorithm and particle swarm optimization algorithm are adopted for generating the energy-optimal trajectories. Simultaneously, the proposed sit-to-stand assistance robot was modeled by SolidWorks, and then importing the geometric model was imported into RecurDyn for further dynamic simulation. Subsequently, the generated energy-optimal trajectories were imported to the dynamic model of sit-to-stand assistance robot in RecurDyn. The optimal trajectory with best combination of time interval set of t_a, t_b, t_c is thus evaluated to be confirmed through RecurDyn simulation. Finally, a four D.O.F.(Degree of freedom) Dobot v.1 arm was selected to form an experiment project to verify the conclusion conducted by the theoretical study. The comparison and discussion of the results solved by the simulation of RecurDyn and experiment verification are concluded.

In conclusion, the solved trajectories under optimization process indeed the optimal results as comparing to the other trajectories. The best energy-saving trajectory providing the dynamic simulation of 3-DOF sit-to-stand assistance robot model and experimental verification using 4-DOF manipulator are all reaching the same conclusion.