Rover continuous path planning using merged perceptions

Abstract

This paper presents a path planning method based on gradient navigation maps merging, for planetary rovers. The trajectories are calculated in continuous mode and perception data are provided by a stereo bench which captures the image pairs during rover movements. Path planning algorithms have been designed and tested using a fully functional robot simulator. A first implementation of the method is being tested with a Marsokhod type chassis equipped with a stereo bench designed to make snapshots when the robot is stopped. We give some details about performances in terms of CPU time and memory space required for such an implementation. Finally, an on going experiment involving our IARES chassis is summarily described.