I am currently working on autonomous navigation and safeguarding of outdoor vehicles, such as an agricultural tractor. Generally, these systems make predictions of upcoming terrain to choose whether it is safe to proceed, but this is difficult in vegetation or slippery terrain where the ground height or friction cannot be sensed directly using forward looking sensors such as cameras or laser range-finders. I am interested in closing the loop around these predictions by letting the vehicle automatically learn the mapping from forward sensor information to quantities of interest such as ground height as the vehicle experiences the world. I am also interested in using probabilistic modeling techniques to improve performance by providing context. See my publications below for more information.
The autonomous tractor project is described more on the following pages, but the information there is quite out of date: NREC project page, NREC project page (old), RI project page, and Tony's project page. I work at the National Robotics Engineering Center and my RI homepage is here.
If you're within CMU, you can view my internal page.