Informed expansion for informative path planning via online distribution learning

Mobile robots are essential tools for gathering knowledge of the environment and monitoring areas interest as well industrial assets. Informative Path Planning methodologies have been successfully applied making able to autonomously acquire information explore unknown surroundings. Rapidly-exploring Information Gathering approaches validated in real-world applications, proving they way go when aiming tasks. In fact, RIG can plan paths with several degrees freedom rapidly complex workspaces by using state-of-the-art Voronoi-biased expansion. Nevertheless, it is an efficient solution most area but its effectiveness decreases exploration/gathering evolves. This paper introduces innovative informed expansion IG tasks that combines Kernel Density Estimation technique a rejection sampling algorithm. By learning online distribution acquired (i.e., discovered map), proposed methodology generates samples unexplored regions workspace, thus steers tree toward promising areas. Realistic simulations experimental campaign, conducted underwater robotics domain, provide proof-of-concept validation developed demonstrate enhances performance