Rapid Subsurface Exploration with Multiple Aerial Robots

Abstract

This thesis develops a robotic exploration framework that allows for rapid and communication-efficient mapping of subsurface environments with a team of aerial robots.

Aerial robots can provide rapid and agile mobility in diverse environments where ground mobility is either severely constrained or impossible. However, high-speed flight with such robots poses challenges due to limited sensing range, on-board computation, and constrained dynamics. For operation in unknown environments, the planning subsystem must guarantee collision-free operation, and for exploration tasks, the system should also select sensing actions to maximize information gain with respect to the environment. To this end, the first contribution of this thesis is a motion primitive-based, receding- horizon planning approach that maximizes information gain, accounts for platform dynamics, and ensures safe operation. Analysis of motions parallel and perpendicular to frontiers given constraints on sensing and dynamics leads to bounds on safe velocities for exploration. These bounds inform the design of the motion primitive approach. Experimental results on a hexarotor robot demonstrate rapid exploration at state-of-the-art speeds in an outdoor environment.

Deploying a team of these robots can further improve the rate of exploration. Challenges imposed by the communication bottlenecks in such deployments towards human-robot and inter-robot coordination have been left largely unaddressed in prior works. Effective coordination often requires high-quality perceptual feedback, and the gap in the state of the art is the lack of efficiency in the communication of such feedback. To this end, the second contribution of this thesis is a distributed perceptual modeling approach that enables high- fidelity mapping while remaining amenable to low-bandwidth communication channels. The approach yields significant gains in exploration rate for multi- robot teams as compared to state-of-the-art approaches. The approach is evaluated through simulation studies and hardware experiments in a wild cave in West Virginia, USA.