The goal of this project is to investigate the potential of fundamentally new modes of robotic manipulation using novel continuum robots in less-structured environments with large uncertainties and even unknowns. A continuum robot, such as a trunk/tentacle robot arm, is in many senses ?dual? to a traditional robot manipulator (consisting of an articulated arm and a gripper or hand), featuring relatively low precision but high compliance:
* Its inherent flexibility and compliance offer greater promise to enable deft manipulation under imprecise and uncertain conditions of objects over a wider range (orders of magnitude) of size and weight, of many different shapes, and with widely different physical characteristics (rigid, soft, flexible, etc.)..
* On the other hand, its inherent lack of precision renders the traditional approaches to planning and control of robot manipulators unsuitable for manipulation with continuum robots.
This project pioneers the study of the basic problem of autonomous manipulation of an object with much uncertainty by a single continuum robot. It introduces a novel and holistic approach that integrates real-time adaptive planning and robust control schemes under real-time sensing and large environmental uncertainty. It next extends the basic approach to address multiple continuum robots working in a common environment, where each robot needs not know the motion of another robot. The research combines theoretical/algorithmic development with real-world validation on an experimental test bed with real trunk/tentacle robots equipped with sensors. The results will be actively disseminated through publications, free software, and real-world demos to impact research, education, and applications.
