On robot skill learning: Self-organizing capability and understanding robot motor capability
Abstract
Recent advances in human-centered robots are driven by the projection that these robots will have a place in our daily life activities as assistive robots. Endowing these robots with the ability of skill learning will enable them to be versatile and skillful in performing various tasks. The problem of transferring human skills to robots raises tremendous research interest in studying human and robot motor skills. Our research aims at investigating a skill representation to make robots autonomous and dexterous in performing various tasks by learning new skills from past skills, and developing a quantitative measure of robot motor capability of a root motor system for the application of transferring human skills to a robot. We first propose a neuro-fuzzy-based, self-organizing skill learning framework to obtain modularity and autonomy of skill representation in a coherent framework. The proposed framework is distinguished by its capability of self-categorizing significant stimulus-response units (SRUs) and self-organizing learned skills composed of SRUs into new skills; it provides robots with the ability of skill learning from task examples instead of manually designed skills. Extensive computer simulations and experiments with a Pioneer 3-DX mobile robot were conducted to illustrate and validate the performance of the proposed neuro-fuzzy-based, self-organizing skill learning framework. We next propose to investigate and develop a quantitative measure, called the pseudo index of motor performance, to measure the robot motor capability in performing a task satisfying the task spatial and temporal constraints. This pseudo index of motor performance is derived from robot kinematics, dynamics, and control with the speed-accuracy constraint taken into consideration. Extensive computer simulations and experimental work on a 6 DOF PUMA 560 robot were performed to validate the performance of the proposed approach in measuring the robot motor capability.
Degree
Ph.D.
Advisors
Lee, Purdue University.
Subject Area
Robotics
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.