\ Abstract— Current operations of planetary rovers, especially the planning and execution of traverse operations, rely on human analysis and estimation of non-geometric hazards based on images captured by the rover. Despite the use of advanced path planning algorithms capable of avoiding obstacles, this limits daily traverse distances. This paper presents a system concept for planetary rovers capable of safe traversal beyond the immediate range of navigation through forward sensing of terrain trafficability, resulting in improved traversal speeds. I. INTRODUCTION The past decades have seen a number of robotic missions to the Martian surface. While these missions have been extremely successful in terms of scientific data gathered, as well as technologies and capabilities demonstrated, the rovers have faced significant difficulty traversing the Martian surface. The most notable example of this is the MER Spirit which was immobilized when one of its wheels was trapped in subsurface sand during a commanded drive in April 2009 – no indication of the hazard was visible while the drive was being planned. Figure 1. Images from MER Spirit. a) Navigation image showing no indication of hazard, b) Wheel embedded in subsurface sand [Photos: NASA / JPL – Caltech] * Research part of the Forward Acquisition of Soil and Terrain data for Exploration Rover (FASTER) project, supported by the European Commission through the SPACE theme of the FP7 Programme, under Grant Agreement 284419. Y. H. Nevatia, J. Gancet, F. Bulens are with Space Applications Services, 1932 Zaventem, Belgium. T. Voegele, R. U. Sonsalla are with DFKI Robotics Innovation Center, 28359 Bremen, Germany. C. M. Saaj, W. A. Lewinger, M. Matthews, F. J. C. Cabrera, Y. Gao are with the Surrey Space Centre, University of Surrey, Surrey, GU2 7XH UK. E. Allouis is with EADS Astrium, Stevenage, SG1 2AS UK. B. Imhof, S. Ransom, L Richter are with Liquifer Systems group, 1020 Vienna, Austria. K. Skocki is with Astri Polska, 00-716 Warsaw, Poland. To reduce the risk of failure, especially immobilization, current concepts for planetary rover operations rely heavily on human involvement and simulation of rover operations. With regards to traverse operations this includes building up a 3D environment of the current surroundings of the rover based on received imagery, identification of hazards including manual identification of regions with suspected subsurface hazards or high slip, and then planning and validation of paths. While suitable for reducing risk by involving experts for scene and terrain analysis, such operations methodologies limit the distance that can be safely traversed each sol to what is in visible range. Continued interest in planetary exploration and the success of recent rovers has led to the planning of several future missions to Mars for the next decade. As the expected scientific return from these missions grow, so do the required capabilities and need for autonomous operations that do not require regular human involvement. One such mission, the Mars Sample Return Mission, would require the rover to traverse a large distance from its landing site to a cached sample, and return with the sample to the landing site within a year. Allowing sufficient time for other required operations such as collecting the cached sample and transferring it to the ascent vehicle, as well as contingencies, results in a required daily traversal of approximately 170m – significantly beyond the capabilities of current operations. This paper presents a system concept enabling planetary rovers to reliably and rapidly traversal of large distances over unknown terrain in preparation for such future missions. The system is based on the forward sampling of soil and terrain characteristics, allowing the autonomous detection of hazards before the rover is at risk. This reduces the need for human intervention and manual analysis of imagery, allowing the traversal to target locations beyond the range of rover sensors. There are three main components of the proposed system: 1. Scout Rover 2. Soil Sensing System 3. Cooperative Autonomy The next section describes the operations concept that is proposed for improved traversal, after which each of these three components is expanded. Finally, a brief overview of the proposed approach for system validation is presented. Improved Traversal for Planetary Rovers through Forward Acquisition of Terrain Trafficability* Yashodhan H. Nevatia, Member, IEEE, Jeremi Gancet, Member, IEEE, Francois Bulens, Thomas Voegele, Roland U. Sonsalla, Chakravarthini M. Saaj, Senior Member, IEEE, William A. Lewinger, Member, IEEE, Marcus Matthews, Francisco J. C. Cabrera, Yang Gao, Senior Member, IEEE, Elie Allouis, Barbara Imhof, Stephen Ransom, Lutz Richter and Krzysztof Skocki a) b)