Active Sampling for Planetary Rover Exploration

Suhit Kodgule
Master's Thesis, Tech. Report, CMU-RI-TR-19-66, August, 2019

Download Publication

Copyright notice: This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.


Planetary Robotics research has expanded beyond simply developing robust navigation strategies for rovers to providing them with the capability of performing intelligent actions so as to develop a better interpretation and understanding of the environment. This will become essential in the future, when rovers explore regions far away from Earth, at distances that would significantly throttle communication with human operators. This research focuses on two problems of interest: Spatio-Spectral Exploration and Active Spectral Reconstruction. For both of these problems we develop a Markov Decision Process framework reliant on using remote sensing measurements that circumvents the partial observability present in these problems. Further, we propose a Monte-Carlo Tree Search based approach for efficiently sampling relevant locations so as to solve these problems. We demonstrate our approach in simulation as well as testing it with the rover Zo{“e} at a Mars-analogous terrain located at Cuprite, NV and highlight its advantages compared to traditional planning strategies.

author = {Suhit Kodgule},
title = {Active Sampling for Planetary Rover Exploration},
year = {2019},
month = {August},
school = {},
address = {Pittsburgh, PA},
number = {CMU-RI-TR-19-66},
keywords = {Robotics, Planetary Exploration, Reinforcement Learning, Path Plnning},
} 2019-08-13T11:07:25-04:00