Press release: Göttingen Scientists Develop Autonomous Walking Robot

Nr. 31/2010 - 16.02.2010

Tiny network enables flexible change of gaits

(pug) Insects can carry out many different movement patterns using their six legs. They use different gaits depending upon whether they are crawling quickly or need to overcome an obstacle. Scientists from Göttingen have now developed a robot that can likewise switch back and forth between several gaits, depending on the environment. The novelty of this is that these different movements are generated from only one circuit network with only a few connections. The scientists aim to build an autonomous robot that functions independently of direct human control. The walking robot was developed by scientists at the University of Göttingen, the Bernstein Centre for Computational Neuroscience and the Max Planck Institute for Dynamics and Self-Organization. The new technology was presented on January 17, 2010 in the online edition of the renowned scientific journal “Nature Physics”.

Repeated movements in humans and animals, such as walking or breathing, are controlled by neuronal units, so-called CPGs (central pattern generators). Scientists have been using this principle in the development of walking robots. Until now, however, every gait of a robot, such as slow walking, running or climbing stairs, required its own CPG, meaning its own control mechanism. The robot received information about its surroundings from different sensors and then selected the CPG that was responsible for the corresponding gait.

The special aspect about the robot from the Göttingen scientists is that it operates on only one single CPG. This enables the robot to learn more easily and faster since it only has one switch point at which information is coordinated. For example: The robot must overcome an inclination using as little energy as possible. As soon as it begins to climb, its energy sensor indicates that the energy consumption is too high. The robot then varies the circuit process between the energy sensor and the CPG until it finds a gait with which it uses less energy. The robot has therefore learned the correlation between incline and movement pattern and can then immediately select the suitable gait in the future.

This functions thanks to the novel CPG, the secret of which lies in the so-called “chaos control”. Without control, the CPG produces a chaotic activity pattern. This, however, can be converted into a constantly recurring pattern that determines the gait of the robot. Depending on the input signal of the sensor, different patterns and therefore different gaits can be created – all, as it were, from one source.

In the future, the Göttingen scientists plan to additionally equip the walking robot with a memory chip. This would enable the robot to carry out a movement to its conclusion even if it is no longer receiving sensory input. If the robot is to overcome an obstacle, for example, then it must take a large step with all six legs consecutively. “Currently, the robot would not be able to handle this task – as soon as the obstacle is out of sight, it no longer knows which gait to use”, says Prof. Dr. Marc Timme from the Max Planck Institute for Dynamics and Self-Organization. “If the robot is equipped with a motorized memory, then he will be able to plan his movements with foresight.”

Note to the editors:
We will be happy to provide photos of the walking robot upon request.

Contact:
Prof. Dr. Marc Timme
Max Planck Institute for Dynamics and Self-Organization
Network Dynamics Group
Bernstein Centre for Computational Neuroscience
Bunsenstraße 10, 37073 D-Göttingen
Phone (0551) 5176-440
E-Mail: timme@nld.ds.mpg.de