A Coordinated Hazardous Source Search algorithm Based on Gaussian Process Regression

Finding the source of a hidden hazardous substance in an unknown environment is of great importance both in military and civil use. This paper presents a coordinated search algorithm that enables robots to perceive the environment and plan the route autonomously. Gaussian Process Regression is applied to estimate the intensity distribution of the hazardous source within the search area. A bi-level nested optimization algorithm, in which the first level optimizes a temporary target and the second level finds the optimal path for each robot to the temporary target, is implemented to search for the hazardous source location via a multi-robot system. A contrast simulation between multi-robot search and single-robot search is carried out for an indoor heat source search scenario, proving the feasibility and effectiveness of the proposed algorithm.