Knowledge Acquisition for a Robot Excavator

The purpose of this paper is to describe the methods currently in use to build a knowledge base capable of intelligently controlling an autonomous robotic excavator. The current aim is that such a robot will be capable of safely and efficiently executing a rangeof excavation tasks with superior performance to that of the human operator. This paper surveys the measures being taken at Lancaster to provide sucha knowledge base and reports on the various approaches in pursuit of particular difficulties concerning the externalising and at present resides in the head of the skilled operator. Whilst this work concentrates on the knowledge base required for robot excavation, the techniques and the problems encountered are relevant to to the whole generic class of intelligent robots which need to be adaptive in response to unpredictable external factors