Computer Vision Talks at Columbia University

Computer Vision for Autonomous Navigation of NASA and DoD Unmanned Vehicles

Larry Matthies

Machine Vision Group, JPL

Wednesday, January/30, 2 PM

Interschool Lab, 7th Floor CEPSR, Schapiro 

Host: Prof. Shree Nayar 

Abstract 

The Machine Vision Group at JPL develops perception systems for autonomous navigation of robotic vehicles for a variety of NASA and DoD missions. I will present an overview of the mission areas we address, the technical problems we are working on currently, and related problems that are candidates for future research. Our NASA work addresses obstacle detection, position estimation, and mapping for Mars rovers, landing hazard detection and precision landing for Mars landers, and landmark recognition for position estimation for orbiters. Our DoD work addresses autonomous navigation of man-portable, urban robots (obstacle avoidance and stair climbing in particular) and day/night, all-weather obstacle detection for off-road autonomous navigation. The last problem, in particular, is leading us to explore a wide variety of active and passive, multispectral sensors for terrain perception under all environmental conditions. I will show highlights of results in each of these areas.