14 DAY TRIAL // It's got to be the Ben Johnson of the robot world, as it rules the race of the two-legged robots. RunBot, developed at Germany's University of G?ttingen can run on two legs, can climb hills and mountains, which it can see using its infrared eyes.
Possessing a gait-altering neural network, the robot does fall occasionally, but it quickly learns to adapt the movements of its two legs to various slopes, which is probably the most important feature. Adaptability is the key to success in nature and RunBot seems to display a lot of it.
"Any walker that can figure out how to adapt to different terrains and slopes is making headway toward an ideal," says Popular Mechanics' Resident Roboticist Daniel H. Wilson. "A robot designer can never predict what problems a robot will have to face, so the best approach is to design the robot to solve new challenges on its own."
The creators of the robot are doing a good job at mimicking human walk, with its two major aspects, the biomechanical design and movement control, performed by an advanced brain pattern. The mechanical aspect has been solved in part by using intelligent hip, knee and torso designs, which make RunBot so fast.
However, it's another of its interesting abilities that allow it to adapt to the environment. The infrared vision records sensory readings as the passes from flat ground to slopes and correlates these readings with earlier experiences, mostly previous falls. The intricate applications integrated to its neural network helps it make the best choices to avoid falling again, so it shifts his center of gravity forward and shortens his steps.
Of course, the robot is not fully autonomous yet, as some support is necessary to make it walk and run. But this is definitely an improvement over other bots, like Honda's ASIMO, that can walk on two feet in a manner resembling human locomotion at no more than to 6 km/h (3.7 mph).
Compared to ASIMO, RunBot is definitely a world class athlete next to a desk clerk.