Autonomous Robotic Map Refinement for Targeted Resolution and Local Accuracy

Abstract

This paper presents a multistage approach to refining the map of an environment and satisfying the targeted resolution and local accuracy by an autonomous mobile robot. The proposed approach consists of two steps. Having a globally accurate coarse map of the environment developed using a conventional technique such as SLAM or SfM with bundle adjustment, the proposed first step plans a path for the robot to revisit the environment while maintaining a desired distance to all occupied regions of interest since the resolution and the local accuracy of the map typically depends on the distance from which objects in the environment are observed. An Unoccupancy Distance Map (UDM) and a reduced-order Travelling Salesman Problem (TSP) techniques are newly proposed to solve this class of problems. In the final step, an online path replanning and map refinement technique is proposed to achieve the targeted resolution and local accuracy of the map. Parametric studies have firstly validated the effectiveness of the proposed two steps. The autonomous capability of the proposed approach has then been demonstrated successfully in its use for a practical mission.