Navegação e controle de robôs móveis cooperativos: uma abordagem baseada em conectividade de grafos

Abstract

This thesis addresses the problem of motion planning and control ofgroups of autonomous mobile robots during cooperative tasks execution. A general framework that transforms several cooperative tasks into the same basic problem is developed thus providing a feasible solution for all of them. The approach enables using a single team of robots to perform numerous different tasks by providing each robot in the team with a single suite of algorithms which are parameterized by the speci¯cities of the tasks. Therefore, even a totally new and unknown task can be executed by the group of robots if this particular task can be transformed to one of the tasks the team is able to execute. The key is the transformation of cooperative problems to individual constrained motion planning problems. In order to do so, the group of robots is modeled as a graph where each robot is a vertex and each edge represents a motion constraint to be satis¯ed. Motion constraints are imposed by the task and by the other robots in the group. These constraints may be used in simple and very well known motion planning techniques in order to plan and control the motion of each individual robot. We present a decentralized solution for the problem, which algorithms are basedon the online modification of pre-specified navigation functions. Examples in sensing, communication and manipulation tasks are presented, eliciting the elegance of the solutions. Finally, experimental results with groups of bothnon-holonomic and holonomic robots are presented.