Learning by ostentation for robotic assembly

Publisher Website
Google Scholar
Abstract
A machine-learning technique is used to enable a robot, consisting of a vision system and a manipulator arm, to learn by observation, and then perform, a variety of block-stacking assembly operations. The robot can perform these tasks in spite of variation of the initial locations of the blocks, imprecision in its movements, and initial uncertainty about its camera location and arm geometry. Important aspects of this work are how continuous phenomena are perceived and mapped into discrete representations, and how the arm and camera are coordinated with each other.

This publication has 0 references indexed in Scilit: