Synthesizing Plant Controllers Using Real-Time Goals

Abstract

This paper introduces a novel planning method for reactive agents. Our planning method handles, in a single framework, issues from AI, control theory, and concurrency that have so far been considered apart. These issues are mostly controllability, safety, bounded liveness, and real time. Our approach is founded on the supervisory control theory and on Metric Temporal Logic (MTL). The highlights of our method consist of a new technique for incrementally checking MTL goal formulas over sequences of states generated by actions and a new method for backtracking during search by taking into account uncontrollable actions. 1 Introduction A reactive plan is a program that specifies the actions to be executed by an agent that continuously reacts to the occurrence of discrete events from its environment. The automatic synthesis of reactive plans is a key component in the design of autonomous systems. It is of interest in AI planning [ Georgeff and Lansky, 1987; Drummond and Bresina, 1990; Dean...