Visibility-based exploration in unknown environment containing structured obstacles

Abstract

How to enable a robot to explore in an unknown environment containing obstacles via local sensing is an important research issue for robot navigation, map building, and localization. In this paper, a reactive and visibility-based exploration algorithm is proposed. By line-of-sight sensors, the robot can find the occlusion points between covered and uncovered regions, and then approach these occlusion points to "sweep" the area. In addition, a prediction model based on local sensing is applied to refine the trajectory when the robot is moving; therefore as the robot tries to cover unknown area, it also tries to maximize the monitoring of covered area. Simulation experiment shows the efficacy of our approach