Robot planning system based on problem solvers

Abstract

This paper is concerned with a planning system for a robot with a hand and an eye which can manipulate blocks. Two different kinds of problem solvers are used. One is mainly composed of a theorem prover based on the resolution principle, and the other consists of a theorem prover based on pattern matching. These problem solvers are called GOAL-FINDER and JOB-SCHEDULER corresponding to their functions. GOAL-FINDER decides a goal state that is suitable for an order of an operator. JOB-SCHEDULER produces a job sequence for a robot to perform the goal state given the constraints of the block world.