CRASH: A Collaborative Aerial-Ground Exploration System Using Hybrid-Frontier Method

Autonomous exploration of unknown environments is a fundamental application in robotics society. In this paper, we propose a novel collaborative exploration framework using a UAV (unmanned aerial vehicle) and a UGV (unmanned ground vehicle). The work is motivated by the wish to combine advantages from different platforms to improve the efficiency in exploration. The ground vehicle carries a long-range laser scanner but travels slowly among obstacles, while the aerial vehicle has a downward-looking stereo camera and flies fast above obstacles. Our algorithm utilizes the complementary features of these two robots and conducts coordinated exploration, while is still flexible that each robot is able to carry out the task independently. In this paper, we combine frontier-based method and motion primitive for local exploration. Also, we adopt a traditional global fail-safe path planning to guide the vehicle to escape local minimum. The proposed framework is implemented on an autonomous collaborative aerial-ground platform. Extensive experiments and benchmarked simulations are conducted to validate the efficiency of the proposed method.