Exploration strategies for building compact maps in unbounded environments
Exploration strategies are an important ingredient for map building with mobile robots. The traditional greedy exploration strategy is not directly applicable in unbounded outdoor environments, because it decides on the robot's actions solely based on the expected information gain and travel cost. As this value can be optimized by driving straight into unexplored space, this behavior often leads to degenerated maps. We propose two different techniques to regularize the value function of the exploration strategy, in order to explore compact areas in outdoor environments. We compare exploration speed and compactness of the maps with and without our extensions.