Hierarchical Scenarios for Behavior Planning in Autonomous Robots

Abstract

Task scheduling and behavior planning are important components for enabling robot autonomy in a number of applied problems. This paper provides a methodology for straightforward description and automatic formation of scenarios that lead a robot or a multi-robot system to a successful completion of a given complex task. We employ a finite-automata approach to definition of individual tasks within the scenario, which as a whole is represented by an oriented graph. A multilevel scenario formation is described for simplification of the traveling salesman problem and for scene analysis in the problem of demolition debris removal. The input of the proposed algorithm is a weakly structured sensory data. The produced output contains the hierarchical scenario in the form of a recursively nested graph.