Searching for regions out of normal conditions using a team of robots

Abstract

Searching for regions in abnormal conditions is a priority in environments susceptible to catastrophes (e.g. forest fires or oil spills). Those disasters usually begin with an small anomaly that may became unsustainable if it is not detected at an early stage. We propose a probabilistic technique to coordinate multiple robots in perimeter searching and tracking, which are fundamental tasks if they are to detect and follow anomalies in an environment. The proposed method is based on a particle filter technique, which uses multiple robots to fuse distributed sensor information and estimate the shape of an anomaly. Complementary sensor fusion is used to coordinate robot navigation and reduce detection time when an anomaly arises. Validation of our approach is obtained both in simulation and with real robots. Five different scenarios were designed to evaluate and compare the efficiency in both exploration and tracking tasks. The results have demonstrated that when compared to state-of-the art methods in the literature, the proposed method is able to search anomalies under uncertainty and reduce the detection time by automatically increasing the number of robots.