Plume Finding with Frontier-Based Planner

Toxic/hazardous gas leaks seriously jeopardize public safety and the ecological environment, and how to utilize drones with on-board gas sensors to quickly detect gas plumes at an early stage of a gas leak and take timely measures can effectively prevent the expansion of the accident. Traditional plume finding methods have the disadvantages of poor targeting, long search times, and repeated exploration in some areas. In view of this, this paper proposes a method of plume finding with frontier-based planner, where a heuristic function is designed to select optimal observation points and thus plan path to search plume. Here an air-mass path model is constructed to estimate the region that would likely be observed by the UAV at a candidate observation site on the frontier, and the area of the unknown part in the region is considered as the environmental exploration gain. The change in flight direction and the distance traveled by the UAV are used as environmental exploration costs when it will fly from its current pose to the candidate observation site. Simulation results show that the proposed frontier-based exploration method for plume finding has good performance in odor source localization tasks in outdoor environments.