Micro aerial vehicles (MAVs) are typically used for automated 3D reconstruction of outdoor environments due to their low cost. However, the use of MAVs for automated reconstruction at the street level and in indoor environments is not straightforward. Given no prior knowledge of the area to be mapped, a MAV has to continually replan a strategy such that it fully covers the area within the shortest time possible, and at the same time, avoid obstacles and walls. In this paper, we describe our approach that makes automated 3D reconstruction possible in such environments. We detail an efficient algorithm for visual exploration and coverage with a micro aerial vehicle (MAV) in unknown environments. This algorithm is designed to run on-board the MAV in real-time. We assume that the MAV is equipped with a forward-looking depth-sensing camera in the form of either a stereo camera or RGB-D camera. We use extensive simulation experiments to benchmark our algorithm against the well-known frontier-based exploration algorithm, and show that our algorithm allows the MAV to fully explore an environment, and at the same time, achieve a significantly higher level of coverage.

@inproceedings{heng15efficient,
	author = {Lionel Heng and Alkis Gotovos and Andreas Krause and Marc Pollefeys},
	booktitle = {Proc. International Conference on Robotics and Automation (ICRA)},
	title = {Efficient Visual Exploration and Coverage with a Micro Aerial Vehicle in Unknown Environments},
	year = {2015}}