Mapping and exploration are important tasks of mobile robots for various applications such as search and rescue, inspection, and surveillance. Unmanned aerial vehicles (UAVs) are more suited for such tasks because they have a large field of view compared to ground robots. Autonomous operation of UAVs is desirable for exploration in unknown environments. In such environments, the UAV must make a map of the environment and simultaneously localize itself in it which is commonly known as the SLAM (simultaneous localization and mapping) problem. This is also required to safely navigate between open spaces, and make informed decisions about the exploration targets. UAVs have physical constraints including limited payload, and are generally equipped with low-spec embedded computational devices and sensors. Therefore, it is often challenging to achieve robust SLAM on UAVs which also affects exploration. In this paper, we present an autonomous exploration of UAVs in completely unknown environments using low cost sensors such as LIDAR and an RGBD camera. A sensor fusion method is proposed to build a dense 3D map of the environment. Multiple images from the scene are geometrically aligned as the UAV explores the environment, and then a frontier exploration technique is used to search for the next target in the mapped area to explore the maximum area possible. The results show that the proposed algorithm can build precise maps even with low-cost sensors, and explore the environment efficiently.
Low-Cost 2-D Map Generation System for a Mobile Robot