Who needs a hiking buddy?
Speedwalking Robot is perfecting its hiking ability
A German scientist from the University of Göttingen developed a walking robot that is able to adapt its gait to the slope of an incline. “RunBot”, so the name of the robot, simulates the neuronal principles which enable humans to adjust their gait when the terrain changes.
According to the scientists, the walking robot uses reflexes driven by peripheral sensors to adjust its movement when the ground changes from flat to slightly uphill. In addition, RunBot features integral control circuits that make sure that the joints are not overstretched. They also guarantee that the next step is initiated not earlier but when the foot touches the ground.
Now when a slope needs to be climbed - the hiking robot knows this by screening the surface with an infrared eye – the built-in circuits adapt to a simpler neural network by changing a couple of parameters.
Florentin Wörgötter, a research scientist at the Bernstein Center for Computational Neuroscience at the University of Göttingen, said that RunBot will fall on its back at its first endeavour to climb a slope. That is because the robot needs yet to learn to react correctly with a change of gait to the visual input through its infrared eye. But after a few trials, RunBot will learn from its failures. Every trial strengthens the connection between infrared eye and movement control, Wörgötter said.
Finally, the hiking robot learns to control its gait length, body posture and off it goes.
***
Update 2016-01-05:
Unfortunately, both, the video and the case study have been removed from their respective publishers. Since the links were dead, I removed them.
Speedwalking Robot is perfecting its hiking ability
A German scientist from the University of Göttingen developed a walking robot that is able to adapt its gait to the slope of an incline. “RunBot”, so the name of the robot, simulates the neuronal principles which enable humans to adjust their gait when the terrain changes.
According to the scientists, the walking robot uses reflexes driven by peripheral sensors to adjust its movement when the ground changes from flat to slightly uphill. In addition, RunBot features integral control circuits that make sure that the joints are not overstretched. They also guarantee that the next step is initiated not earlier but when the foot touches the ground.
Now when a slope needs to be climbed - the hiking robot knows this by screening the surface with an infrared eye – the built-in circuits adapt to a simpler neural network by changing a couple of parameters.
Florentin Wörgötter, a research scientist at the Bernstein Center for Computational Neuroscience at the University of Göttingen, said that RunBot will fall on its back at its first endeavour to climb a slope. That is because the robot needs yet to learn to react correctly with a change of gait to the visual input through its infrared eye. But after a few trials, RunBot will learn from its failures. Every trial strengthens the connection between infrared eye and movement control, Wörgötter said.
Finally, the hiking robot learns to control its gait length, body posture and off it goes.
***
Update 2016-01-05:
Unfortunately, both, the video and the case study have been removed from their respective publishers. Since the links were dead, I removed them.
Comments
Post a Comment