USC Robotics Lab. Logo

Videos of the USC Robotics Lab's Segway RMP

These videos may not be used for commercial or for-profit purposes. They are copyrighted by the authors.

Here are some videos of the USC Robotics Lab's Segway RMP driving around, using the Player robot device server. Look here for information on how to do this yourself (assuming that you have an RMP handy).

For videos of a simulated Segway RMP running about in Gazebo, look here.


These experiments were carried out on 21 June 2003 by Brian Gerkey and Andrew Howard.

Thanks to John Sweeney for writing the original Segway driver and to Chris Jones for implementing Johann Borenstein's Vector Field Histogram (VFH) algorithm as a Player driver; we used it throughout the autonomous tests shown below.

All videos are in real time, formatted as AVIs, using the DivX codec.

First drive off the lot. This is joystick control, using the Player utility playerjoy.
[3.2MB]
Autonomous wall-following, at 1 meter/second (fixed camera angle).
[3.7MB]
Autonomous wall-following, at 1.5 meter/second (camera following robot).
[3.0MB]
Autonomous wall-following, at 1.5 meter/second (camera strapped to robot).
[4.8MB]
Outdoor autonomous obstacle slalom (head-on angle).
[2.8MB]
Outdoor, off-road, joystick control.
[4.4MB]
Outdoor, spectacular fall under joystick control (follows previous clip).
[1.2MB]

These experiments were carried out August 2003 by Andrew Howard, Boyoon Jung and a cast of thousands.

Thanks to the folks at the UPenn Grasp Lab for the formation control equations used in all of these videos.

All videos formatted as QuickTime, using the DivX codec.

Target following at Fort Benning (around the Triple-Wide)
[14MB]
Target following at Fort Benning (long walk up to the MOUT site)
[10MB]
RMP's in formation: Georgia Tech in the lead (tele-op), UPenn in middle (visual tracking), USC in the rear (laser-based tracking).
[23MB]