UK researchers to give legs to robot vision

Wheeled robotic systems which navigate with the help of vision systems have been widely used over the past few years for numerous tasks including planetary exploration and surveillance. But while they can be fast and efficient on hard flat ground, the natural legs of animals are more versatile and efficient on natural terrain.

Now, researchers at Bristol University (Bristol, UK) led by Dr. Jeremy Burn are involved in a £500,000 project sponsored by the Engineering and Physical Sciences Research Council (EPSRC) that will see them create a vision-based control system for legged robotic systems that can travel over complex terrains.

To do so, they plan to capture visual scenes of terrain using a head-mounted camera. A map of the terrain synthesized in 3-D virtual space and updated in real-time will then be presented to a human subject using a virtual reality headset. By doing so, the researchers will then be able to investigate how the gait and limb movements of the subjects to compensate for partial or unreliable information about the environment.

The research will also provide an insight into how feedforward and feedback control systems could be developed for such legged systems, as well as allow the researchers to determine the performance it would be possible to obtain from a vision-based legged robotic system.

More details on the three-year research program, which runs from July 2012 to July 2015, can be found here.

-- by Dave Wilson, Senior Editor, Vision Systems Design