Point Cloud Based Autonomous Area Exploration Algorithm

Abstract

Autonomous navigation is highly important in robotics, especially when it comes to the robotic applications in disaster management etc. There are many algorithms to implement autonomous navigation and most of them are dependent on prior knowledge of the environment and apriori maps. Although they are effective in some scenarios, these algorithms fail to perform when the environment has been subjected to changes that might invalidate the prior map. This paper presents a point cloud based algorithm which can be used in a situation where the prior knowledge of the environment is highly inaccurate. The proposed algorithm uses depth images to get a local map, which it expands by searching for uncharted areas picking the next best location to explore using a breadth first approach given a set of constraints. The proposed algorithm exploits the maps in the 3D space allowing the navigation system to perform effectively in uneven terrains and use inclined planes for its advantage.