Development of an Autonomous Robot for Ground Penetrating Radar Surveys of Polar Ice Eric Trautmann, Laura Ray, Jim Lever Abstract— This paper describes the design and fabrication of a low cost, battery-powered mobile robot for ground penetrat- ing radar surveys in support of Polar science and logistics. Key features of the design include lightweight construction for low resistance and high energy efficiency in deformable terrain; a passive, articulated chassis for high mobility; and design simplicity for low cost. Deployment in Greenland in spring 2008 over crevasse fields demonstrated the ability of the robot to traverse rough terrain characterized by both firm and soft snow, while gathering data from a ground penetrating radar to detect crevasses. A simple navigation and control algorithm provides low-bandwidth path planning and course correction. Mobility assessment during deployment highlights the need for non-visual means of assessing mobility autonomously. A proprioceptive sensor suite and sample data for autonomous detection of terrain traversability are described. I. INTRODUCTION Recent increases in scientific research in the Antarctic and the Arctic have led to greater demands for logistic support services. Remote scientific bases like the South Pole Station, and Greenland’s North Eemian Ice drilling (NEEM) and Summit stations have historically been resupplied by air, though recent logistics efforts have focused on developing overland resupply as a more economically and environ- mentally efficient solution. Sub-surface crevasses, caused by shear zones in moving ice sheets, pose a serious danger to both personnel and equipment. Traverse teams currently use a Ground Penetrating Radar (GPR) unit, suspended on a boom from the front of a tracked Sno-Cat vehicle as shown in Fig. 1, to image terrain in front of the vehicle. This method gives the vehicle operator approximately two seconds of warning time to stop the vehicle after detecting a crevasse. In this paper we present the Yeti robot, a robotic platform to perform autonomous GPR surveys of polar ice. This system was designed for the purpose of crevasse detection, but could also be used for GPR-based scientific research such as meteorite detection [1] or ice stratigraphy [2]. Yeti is a four- wheeled, differentially-steered autonomous robot equipped with a proprioceptic sensor suite and a SIR-3000 ground penetrating radar unit from Geophysical Survey Systems Inc. This work was supported by the Jet Propulsion Laboratory under Grant No. RSA1310519, the Army Research Office under contract No. W911NF- 06-1-0153, the National Science Foundation, and the U.S. Army Cold Regions Research and Engineering Laboratory. E. Trautmann is a Masters Student of Electrical Engineering, Dartmouth Thayer School of Engineering etrautmann@dartmouth.edu L. Ray is faculty of Mechanical Engineering and Roboftics at Dartmouth College, Hanover, NH 03755 laura.ray@dartmouth.edu J. Lever is a researcher with the Cold Regions Research and Engineering Laboratory, United States Army. Hanover, NH 03755 James.H.Lever@erdc.usace.army.mil Fig. 1. The Yeti robot next to the Tucker Sno-Cat. Yeti will supplement or replace the Sno-Cat for performing Ground Penetrating Radar Surveys to detect sub-surface crevasses in polar ice sheets. A. Related Work Current crevasse detection methods use ground penetrating radar to isolate areas of ice with characteristic low radar reflectivity profiles [3]. The traverse teams operating in the Arctic and Antarctic deploy a radar unit from a Sno-Cat. This platform is expensive ($500,000+), slow in rough terrain, and dangerous for the operators [4]. Several alternative human-operated GPR survey tools have been developed. In February 2009, the Moon Regan Antarc- tic traverse will use the Concept Ice Vehicle, a propeller- powered vehicle on skis, to deploy a GPR unit in front of the traverse team. The CIV, designed by the automaker Lotus, is optimized for high speed at the cost of maneuverability over rough terrain [5]. Several autonomous robotic platforms have been devel- oped to traverse polar terrain. Researchers at Carnegie Mellon University developed the NOMAD robot, a four- wheeled, GPR-equipped vehicle in collaboration with NASA, to autonomously search for meteorites on antarctic terrain. NOMAD has a mass of 725 kg and is about the size of a small SUV, in large part addressing the challenge of negotiating terrain features with its large size [6]. This design comes at the cost of higher power requirements, greater ground pressure, and lower safety on snow bridges covering sub-surface crevasses. The Cool robot, designed at Dartmouth College to be used in teams to deploy networked sensors on a mobile platform The 2009 IEEE/RSJ International Conference on Intelligent Robots and Systems October 11-15, 2009 St. Louis, USA 978-1-4244-3804-4/09/$25.00 ©2009 IEEE 1685