Solving Problems in Robotics with Semantic Networks

Abstract

Robot problems are examined in the context of semantic networks which are used to represent the state of a problem and the operators useful for solving it. Graph transformation algorithms are discussed as an aid to problem solving. Although these form only a small subset of the first-order predicate calculus based systems, considerations such as subgoal circularity, partially specified states and multiple manipulators sharing the same environment may warrant this simplification.