|
Retired Robots
|
|
|
|
 |
The Frankie project
The Frankie project seeks to build a state-of-the-art mobile robot
using very low cost technology.
Current status (April 13 1994)
At present, the Frankie system is a limited reimplementation of Polly, a low
cost vision-based robot that gives primitive tours of our lab. The
current Frankie system implements the collision-avoidance algorithm
from Polly on a
$700 active vision system. The algorithm is perhaps the most
exhaustively tested vision navigation algorithm to date, having been
demonstrated at such diverse locations as the 1993 AAAI conference and
West Point.
We are presently developing support hardware and software for the
system:
Long-term goals
Our goal is to have Frankie not only give tours, but also read its
email and answer simple questions about the lab over the web. Frankie
will use radio modems both to log into the local Unix cluster and to
upload slow-scan video to the internet.
Researchers
- Ian Horswill,
<ian@ai.mit.edu> (software)
- Chris Barnhart,
<ceb@ai.mit.edu> (hardware)
- Rob Gruhl <morpheus@athena.mit.edu>
- Kinu Masaki <kinu@ai.mit.edu>
- Tito Pena <tpena@ai.mit.edu>
- Cat Tait <cat@ai.mit.ed>
Watch this space
Ian Horswill --- ian@ai.mit.edu
A new stereo vision system with 30 times more compute power.
A $100 robot arm.
A port of the Cornell Mobot Scheme system.