結合Kinect視覺與力輔助策略之機器手臂操控系統

Abstract

在機器人的應用中使用於工廠的機器人為其主流，這類的機器人多半藉著高精準度的環境規劃，並依照工作種類不同，詳細安排工作流程，以達到快速、量產的目的，同時更能在日漸趨於老年化的社會當中，解決勞動力不足的問題。近年來，另一種機器人的運用，則是擔任助手的角色，相較於前者來說，在執行任務時，仍是由人來操作，機器人則是從旁負責輔助的工作，這樣的好處是可降低失誤的發生，並提升工作效率，常見於危險且無法事先進行規劃的場所，如核災現場、高風險的生化實驗室、外科的手術台等，本實驗室致力於開發此類型的機器人操作系統，並針對不同的適應性工作，提供適合的輔助策略，改善操作者的工作效率。本論文的重點為利用本實驗室所開發的基於六軸力回饋搖桿之六軸機器手臂操控系統，並設計數種力輔助策略，以因應不同工作需求，此外，也裝置Kinect視覺系統在機器人端作為視覺定位使用，除了讓系統即時偵測環境的突發狀況，修正目標物件的座標，其深度資訊也可供操作者消弭視差的影響，再透過力搖桿的輸出力將距離資訊回饋給操作者，進而達到即時輔助和遠端操作的目的，經實驗證明此系統具實用性。

The applications of robots nowadays are mainly in factories, not only solving the labor shortage problem, but also increasing production efficiency. However, their usage highly depends on high-precision environmental planning in accordance with the type of works, and they are usually unable to cope with unknown environments. Recently, the assistive robot systems are introduced, which help the human to manipulate the robot by providing certain tools in real-time. Compared with the former one, it is more suitable for unknown environments such as nuclear disaster site and biochemistry laboratory. The aim of this thesis is to deal with the compliance tasks by utilizing the six-DoF force-relation joystick to manipulate six-DoF robot manipulator, along with the proposed force-assistive strategies. The system computes the force information from the force sensor and target position information using the Kinect visual device to generate assistive force for guidance. Experimental results show that this developed manipulation system can enhance the effectiveness during operation, so that natural mmanipulation is achieved.