The University of Edinburgh -
Division of Informatics
Forrest Hill & 80 South Bridge

MSc Thesis #92131

Title:Sensorless Soma Part Orientation
Date: 1992
Abstract:In assembly robotics, a problem is acquiring parts in accurate positions and orientations. This project focuses on the mechanics of pushing and how knowledge of pushing can be applied in the automatic planning of orientation movements. The planning problem is to create a sequence of effective sweeping motions which will align a part by minimising the uncertainty in the part's orientation. The initial orientation of the part is completely unknown. The objective of this project is to devise a set of sweeping behaviours which will properly orient each soma4 part despite rotational uncertainty. These push sequences were then implemented and tested with the Adept1 robot arm. This report details the development, design and implementation of the system to plan push sequences to recover the soma parts from a random initial orientation to a known orientation. The planner has been implemented and tested on the Adept1 robot. This technique adopted the push-stability diagram of Brost [R.C.85], [R.C. 88] and the representation of the qualitative behaviour of a sliding workpiece came from Mason's PhD results. We describe how this diagram can be used to plan how to put a planar workpiece into a known orientation from an unknown random orientation.

[Search These Pages] [DAI Home Page] [Comment]