Title: Detecting and predicting environmental boundaries with a team of robots

Environmental boundaries, such as the borderline of a forest fire or an oil spill, pose a significant threat for living organisms. Anticipating the dynamics of these phenomena is a potentially life-saving indicator to support efficient and effective evacuations or to dispel the hazard. We propose a decentralized coordination method that allows multiple robots to efficiently sample and predict the behavior of environmental boundaries. Our method does not require a priori information about the boundary dynamics. We validate our proposal through experiments with actual robots. We show how the accuracy of our estimation method is affected by different factors. We demonstrate experimentally that our method can estimate and predict non-convex boundaries even with noisy measurements and inaccurate actuators.

Prof. Mario Fernando Montenegro Campos – Advisor (DCC – UFMG)
Prof. Renato Martins Assunção – Co-advisor (DCC – UFMG)
Prof. Ani Hsieh (Drexel University)
Prof. Luciano Cunha de Araújo Pimenta (DEE – UFMG)
Prof. Luiz Chaimowicz (DCC – UFMG)
Prof. Rafael Fierro (University of New México)