A new paper by David Maiwand of the CMU Robotics Institute describes a vision-based localization and system identification method use with the RHex hexapod robot. System identification is the process of fitting a mathematical model to the visually observed data. The system relies on a YUV422 320x240 pixel camera. The system was implemented in C++. The paper describes the hardware, software, and algorithm in great detail. Some of the related source code can be found on David's website.


del.icio.us
Digg
Google bookmark
reddit
Simpy
StumbleUpon
Furl
Newsvine
Technorati
Tailrank