Home Research Sections Section 3
Personal tools

Section 3

Adaptive Neuronal Control Technology

Adaptive neuronal control architectures and algorithms: The domain of neuronal control algorithms has experienced several remarkable developments in the last years mainly in the fields of prosthetics and robotics. Only recently it has become possible to control humanoid robots with many degrees of freedom by biologically realistic neural networks that can adapt and learn. In general, one finds that neuronal control is much more flexible and robust than conventional (PIDlike) control methods. Major achievements exist here for example in imitation learning by simulated mirror neurons and in dynamic, adaptive limb- and walking control. Similar techniques are being investigated for prosthetic devices. Furthermore, the theory of closed-loop control by biological networks has made progress through advanced models of micro-circuitry a better theoretical understanding of human central sensory-motor control and motor programming and the attempts to transfer this to machines.

Projects involved:

3b - Control of multi-joint multi-sensor hand prostheses

J. Michael Herrmann, Armin Biess, Florentin Wörgötter, and Otto Bock HealthCare GmbH