Collaborative robotics and its paradigms for industrial manipulators

• Main Author: Weng, Ching-Yen
• Other Authors: Chen I-Ming
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Industrial engineering; Engineering::Electrical and electronic engineering::Control and instrumentation::Robotics
• Online Access: https://hdl.handle.net/10356/145983
• Institution: Nanyang Technological University

Industrial robots can be applied to a wide range of feasible applications through various levels of programming and integration into manufacturing systems. By the cooperation between various agents, various collaborative patterns can be introduced to achieve their industrial objectives and improve many aspects of the task performance, such as time-efficiency, dexterity, and flexibility. Although this concept has been largely studied, it is too novel to be systematically defined in recent studies, especially for the use cases of industrial applications, leading to the difficulties of mature continuous integration and continuous delivery in manufacturing. To enable the design of industrial manipulators to be collaborative with other agents, such as other robotic systems or humans, a promising method is expected to turn such an abstract concept into a mathematical model so that the existed technologies of performance assessment can be applied to evaluate the deployment of systems. In addition, to consider both the precision and flexibility in human-robot collaboration for the industrial manipulators, to propose a unified framework to cover the contents of collaborative manipulation would support a more effective task execution with human-robot collaboration. To make up for the gaps in research and industrial requirements, this thesis studies collaborative robotics in the applications of industrial manipulation. Three novel aspects of the problems in the field of collaborative robotics are studied: ``the definitions and classifications'', ``task representation and performance assessment'', and ``collaborative manipulation''. In the first aspect, commonly used terminologies are clearly defined to avoid confusion and a general classification is proposed to group various collaborative patterns and forms. In the second aspect, a task representation method is developed to conform to the aforementioned classification. Furthermore, this representation is able to model robotic tasks for performance assessment, which can provide a more scientific manner to develop and deploy a robotic system in manufacturing. In the third aspect, a collaborative manipulation framework enables human operators to collaborate with robots in a more flexible, intuitive, and time-efficient manner. To validate the effectiveness, a vision-based peg-in-hole task is conducted in a robot-robot collaborative pattern for the performance assessment. Besides, an aerospace masking task is improved via the telemanipulation with a human-robot collaborative pattern. These two cases effectively demonstrate the applications of the developed theories and prove the feasibility of the proposed methods in this thesis, which not only imply the theorizability of collaborative robotics but also show its potential values in more industrial applications.