俱多模式直覺教導-執行功能之七自由度冗餘機器手臂系統

Abstract

近年來，由於勞力成本的提升，產業自動化的需求也跟著增加，但對於產線變化大的應用，人工仍比機器人更具彈性與智慧，也是較有效率的選擇。這其中需要的技術突破是全面提升機器人的易使用性與機器人的智慧自主能力，而提升彈性與易使用性則是近程可以達到的目標。已有研究提出機器人的示範教導功能(teach by demonstration)來提升易使用性，意即使用者直接接觸機器人並推著或拉著機器人完成動作，此後機器人便可重複執行該示範的動作。而許多方式是需要由力感測器或關節扭力感測器的回授來達成，而且只有單純的教導功能。 因此，本論文提出與實作了一種可用於七自由度冗餘機械手臂，並且具有五種功能的多模式直覺式教導系統，而且沒有使用力傳感器。此五種功能增加了教學系統的實用性和直覺性，包括1）固定機械手臂末端方向與自運動的教導，2）固定末端位置的教導，3）固定末端方向與位置的教導(純自運動的教導)，4）末端直線運動的教導，和5）末端平面運動的教導。此功能的靈感源自於我們發展整個七自由度機器人系統的經驗，主要包括七自由度機器人的逆運動學分析與冗餘阻抗控制。因此，本論文由整個七自由度機器人系統架構來論述，從運動學分析、軌跡規劃、運動控制到主題的教導功能。在逆運動學方面，除了本論文主要的七自由度手臂的分析，另外提出了兩種關節組態的逆運動學解析解，此組態在文獻中是缺少的；在軌跡規劃方面，提出了一種基於正規化向量計算的線上軌跡產生器，可達到多軸運動相位同步與空間速度、加速度限制的功能，使用線上軌跡產生器於手臂控制架構中使其他高層規劃功能的整合變得更簡單，另外，在實用的路徑命令功能方面，實作了一種基於線上軌跡產生器算法的路徑插補器；在控制系統方面，除了基本的多軸伺服控制，也提出一種基於冗餘參數的自運動控制方法於阻抗控制架構中；最後呈現了多模式直覺教導功能的實現方法，結合順應關節、運動學解偶、冗餘阻抗控制等技術整合於控制系統架構中，以實驗結果檢驗其功能性，並且展現一應用示範：抓取與放置(pick & place)的示範。

In recent years, because of the increasing labor costs, requirement of industrial automation also increase. However, for the application in the production line with big variation, human still have more flexibility, intelligent, and efficiency than robot. The technological breakthrough needs robot more and more easy to use and with more and more autonomous intelligence capability. The former is a short-range goal can be achieved. Some methods of “teach by demonstration” have been proposed to increase easiness of usability. It means users direct contact with robot and pushing or pulling the robot to complete the action, then the robot can repeat the action of the demonstration. Many methods needs force sensor or joint torque sensor feedback. Therefore, this thesis proposes and implements a 7-DoF redundant robot manipulator multimodal intuitive teaching functions, which have five teaching functions, and without force sensor. The five functions increase the utility and intuition of the teaching function, which include 1) guide with fixed orientation, 2) guide with fixed position, 3) guide with null space self-motion, 4) guide with linear motion constraint, and 5) guide with plane motion constraint. To implement this work, firstly, the kinematic analysis of the 7-DoF robot has been derived with a specific redundant parameter. Second, the necessary impedance control scheme is achieved and a method of the null space self-motion control based on the redundant parameter is proposed. Third, the intuitive teaching system and multimodal guidance are developed, the method combine the soft joint control, kinematic decoupling and redundant impedance control, and presented and integrated. Finally, experimental demonstration results for verifying the functionality of the developed system on our own 7-DoF robot manipulator are presented. Furthermore, an application demo “pick & place” is presented at last.