Vision-based following for cooperative astronaut-robot operations

Autonomous robots will provide crucial support for astronauts aboard the space station, during extravehicular activities (EVA), and on planetary surfaces. To be useful, a robotic assistant should be capable of autonomously executing commands, safely following an astronaut, and operating under human control. In this paper, we outline an adjustable autonomy architecture for such an assistant and describe our efforts toward robust astronaut following during free-flight. We have applied a color-based vision tracking system to the astronaut following task on the SCAMP Space Simulation Vehicle (SSV). This paper describes SCAMP SSV, its onboard control system, and the vision-based tracking system. We discuss initial test results to quantify tracking performance and follow-on research to realize the adjustable autonomy behavior we have proposed.