Controle de formação de robôs móveis baseado em visão omnidirecional

Abstract

This work addresses the problem of controlling the formation of a group of mobile robots. A centralized control was chosen and projected according to nonlinear control theory. The team is composed by a leader with major processing capacity, and other simple and cheap robots, called followers. The leader is responsible for driving the group to the desired formation during navigation and is equiped with an omnidirectional vision system in order to estimate the followers position and orientation by capturing just one image. The feedback of the controller is totally based on vision and is obtained through digital image processing techniques. The algorithms used to process the omnidirectional images improve the system robustness and performance. Simulations were carried out in order to study the controller behaviour facing different group formations. The results obtained through the experiments with real robots show that the proposed control system has achieved a good performance.