Implementação de algoritmo de visão computacional para comunicação entre membros de um enxame de robôs

Abstract

Systems with multiple robots, also called swarm robots, are growing and getting more importance in research. Due to the large number of applications, the drones are leading this growth, being used for exploitation of environments, filming and deliveries. An essential requirement for multi-robot systems behave properly is the communication. As the conventional way of communication can be under various kind of failures, emerges as an alternative the communication using computer vision. Thus, inspired on bees visual's communication, the present work aims to identify objects of interest in the drone's field of vision and analyzes its movement, in order to recognize what information that object is sending. Hence, were implemented the Lucas-Kanade's algorithms and Harris's corners detection, with the purpose of segmentation of these objects of interest. At last, template matching techniques were used to identify expected models, acquiring satisfactory results on information's recognition transmitted visually, which achieved the level of 75% of success.