Publikation

A Bio-inspired Control System for a Wearable Human-Machine Interface

Michele Folgheraiter, Mathias Jordan, Jan Albiez, Frank Kirchner

In: International Conference on Adaptive Behaviour. International Conference on Adaptive Behaviour (SAB-12) August 27-31 Odense Denmark 8/2012.

Abstrakt

A control system is presented that integrates bio-inspired with classical control techniques to govern the forearm joint of a wearable haptic interface. The neural circuit is based on the architecture of the human segmental reflexes, and the neurons are represented by the combination of a first order linear differential equation and a sigmoid or a piecewise-linear activation function. Due to a long term adaptive mechanism that considers the state of the interface and the interaction force with the user, the stiffness of the joint is regulated according to the particular motion and task at hand. Experimental results showed that the proposed control architecture is able to improve the interface performances in terms of responsiveness, as well as to implement a safety behavior that intervenes in case of harmful external forces.

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence