Uluç Saranli's Research: Dynamic Self-Righting
Overview Publications Software Readings Researchers

Recovery of proper body orientation is one of the most basic robotic self-manipulation tasks. It becomes most critical in autonomous scenarios as well as teleoperation, where the robot may be unreachable. In such cases, the inability to autonomously recover from even a simple fall may render the robot completely useless, decreasing the overall robustness of the platform for real-world missions.

Traditionally, robots adopt specialized kinematic structures and actuators specifically designed for relf-righting. RHex is not equipped with such mechanisms and cannot achieve self-righting through purely kinematic maneuvers. RHex can only achieve self-righting by use of its body dynamics to manipulate its orientation. We have recently designed a feedback controller based on a planar model of RHex that can dynamically perform back-flips over a variety of indoor and outdoor surfaces such as carpet, asphalt and grass. This controller serves both as a useful behavioral tool and as an important step towards more sensor based dynamic behaviors on RHex.

Movies
Open-loop controller on linoleum
Open-loop controller on carpet
Model based controller on linoleum (150fps)
Model based controller on asphalt
Model based controller on dirt
Model based controller on grass
Check out RHex's project site for more movies.
Relevant Publications
Related Links
created by Uluc Saranli, last modified
[ Home ]