3D surface mapping using a semi-autonomous rover: A planetary analog field experiment

This paper describes a proposed operational architecture for a planetary worksite mapping mission concept. To map three-dimensional (3D) planetary terrain, we propose to use a rover equipped with a laser rangefinder, and employ a stop-scan-go approach with a human-in-the-loop. In the operational cycle, the rover collects locally consistent 3D range data while stationary. The range data are coupled with visual odometry to estimate the rover pose at each scan and create a consistent 3D map. The 3D map is then used to evaluate candidate next-best views (NBV). The operator selects a NBV with the aid of three evaluation criteria and the rover autonomously travels to the NBV using a network of reusable paths (NRP). Finally, the rover collects another 3D scan and the cycle repeats. This mission concept was validated through hardware experiments on the CSA’s Mars Emulation Terrain (MET), which measures 60m x 120m and includes inclines, rocks, cliffs and a 5.5m-diameter crater.