Skip to content Skip to navigation

SLAB’s Cutting Edge Research on Spacecraft Pose Estimation Finds New Applications with Autonomous Quadcopters

By Josh Sullivan   May 9, 2016

SLAB’s very own Ph.D. candidate, Sumant Sharma, has teamed up with three other Stanford graduate students to rise to the demands of this year’s exciting DJI Developer Challenge.  DJI has partnered with Ford and the United Nations to create a unique competition in which teams must demonstrate advanced quadcopter search and rescue systems.  Competition tasks require the quadcopter to autonomously navigate and survey a “disaster area,” transmit visual data on potential “survivor” locations back to a moving vehicle, and automatically rendezvous and land on the moving vehicle.  The Cardinal S.A.R team was selected from a highly competitive pool of worldwide challengers to secure one of the 25 coveted competition spots and obtain designated DJI developer hardware.

While Sharma’s research efforts have been largely focused on improving spaceborne pose estimation capabilities, he is now leveraging this work in terrestrial applications to enhance the ability of the Cardinal S.A.R quadcopter to navigate using camera-based measurements.  The intent is to demonstrate the generality of the cutting edge close-range pose estimation algorithms in a new and previously-untested scenario.  Speaking on his motivation for joining the team and applying his research to the challenge, Sharma said: “It’s giving us an opportunity to rapidly prototype and test the algorithms on real hardware- a graduation from the purely simulation-based testing I’ve been conducting.”  The pose estimation algorithms enable the onboard navigation system to determine the position and orientation of an observed target relative to the surveying quadcopter solely from images taken by the onboard camera.  While the DJI challenge takes place in a vastly different environment than space, there are clear similarities in the algorithm implementation.  For example, the approach and landing of the quadcopter on a moving vehicle marked with “April Tags” is a reasonable terrestrial analog to the autonomous rendezvous and docking of two spacecraft making use of fiducial markers.

The advancement of spacecraft Guidance, Navigation, and Control (GN&C) technology is a fundamental research pillar of the Space Rendezvous Lab.  This competition, however, highlights the fact that the innovative research coming out of SLAB has multi-faceted applications for distributed systems in general.  While our research continually paves the way to a new chapter in space technology and exploration, it’s exciting to take our eyes from the stars and see that path extending to improvements back on the ground.  



Josh Sullivan is a graduate student in Stanford’s Space Rendezvous Lab