A Multiple Information Source Planner for Autonomous Planetary Exploration

Abstract

This research casts the problem of exploration into one of collecting information and enables the completion of complex exploration tasks by considering multiple sources of information. This paper describes a methodology to solve exploration tasks which computes the expected information gain, for each information source, from a sensor reading. It also proposes a greedy search algorithm to collect information while accounting for costs such as driving, sensing and planning. An instantiation of the methodology to solve the exploration task of creating traversability maps is presented. This task uses three information sources: frontier detection, increase map certainty and determine reachability. These sources are explained in detail. Finally, the traversability map creation task is performed in simulation.