Exploration of an Unknown 2D Environment using a View Improvement Strategy

In this paper we present a new exploration strategy for a mobile robot moving in a 2D plane. The task of the exploration algorithm is to move around autonomously in an unknown room environment of a limited size whilst building and using a 2D map to avoid obstacles and plan its next move. The strategy is not goal driven, rather seeking new unseen areas to view and explore. The exploration strategy is designed for a vision system with a limited field-of-view. Improvement from previous views of local obstacles is used with a next best viewpoint calculation to select the next move where recently detected obstacles are given more weight. Example explorations are presented and are shown to find new areas to view without exhaustive searching.