Leader-Following for Mobile Robots

TRACLabs Inc. has worked with NASA to create indoor/outdoor, vision-based leader-following systems for sparse, extra-terrestrial environments. Brown University's Robotics Group has demonstrated leader-following and gesture recognition in high-traffic, dynamic, indoor environments using active lidar sensing. We propose to combine our technologies in order to create a passive sensing system that will facilitate unmanned leader-following in novel, cluttered, and dynamic environments. We plan to accomplish this goal via the synthesis of TRACLabs' technologies for stereo-vision-based pursuit, safe 6-degree-of-freedom navigation, and multi-sensor object tracking with the Brown Robotics Group's innovative algorithms for gesture recognition and person-following. Our Phase I aims are 1) to demonstrate an initial leader-following prototype, using our existing technologies to achieve close-range, vision-based leader-following in cluttered and dynamic indoor/outdoor environments and 2) to provide a detailed assessment of the hardware and software extensions needed for the long-range Phase II system. In Phase II, we plan extend our system to long-range outdoor leader-following, including conveys of non-human leaders. We will focus on leader-following at human running speeds and on utilizing alternative passive sensing, like thermal sensing, to achieve robust tracking at long-range distances. Long-term, we expect to integrate our system with unmanned military robots, like PackBot, Talon, and MULE platforms.