Multi-robot planning : a timed automata approach

Abstract

This paper describes how a network of interacting timed automata can be used to model, analyze, and verify motion planning problems in a scenario with multiple robotic vehicles. The method presupposes an infrastructure of robots with feedback controllers obeying simple restriction on a planar grid. The automata formalism merely presents a high-level model of environment, robots and control, but allows composition and formal symbolic reasoning about coordinated solutions. Composition is achieved through synchronization, and the verification software UPPAAL is used for a symbolic verification against specification requirements formulated in computational tree logic (CTL). In this way, all feasible trajectories that satisfy specifications and which moves the robots from a set of initial positions to a set of desired goal positions may be algorithmically analyzed. The trajectories can then subsequently be used as a high-level motion plan for the robots. This paper reports on the timed automata framework, results of two verification experiments, promise of the approach, and gives a perspective for future research.