Efficiency and flexibility of biological systems is still unreached in current robotic systems. Biological evolution formed highly specialized systems, perfectly designed with respect to material, force-to-weight ratio and energy turnover. This can be seen in the human hand, the musculoskeletal construction of which has only recently been understood in all its detail.

Although technical hands, such as the highly integrated DLR Four-Finger Hand, can already be used in a large range of applications, such hands are far from offering an alternative to the human hand, because of size and flexibility. Furthermore, a sensor with properties close to those of the human skin is far from being available to robotic hands.

To reach the same dexterity as the human hand, our solution is to construct a precise kinematic model of the human hand using in vivo MRT (magnetic resonance tomography)-data and constructing a model and robotic hand-arm system from that. The resulting robot hand will be very human-like, and can therefore be optimally connected to and controlled by the human peripheral and central nervous systems.

To ensure an optimal connection between robot and human, we investigate various interfaces:
Non-invasive: We concentrate on electromyography (EMG), placing electrodes on the skin to measure muscular activity. This approach is ideally suited for, e.g., active hand prostheses;
Invasive: We investigate a connection to the human peripheral nervous system (PNS) by inserting electrodes into nerve fibres.

In order to deliver sensory data back to its operator we develop an artificial skin-like touch sensor, based on properties of the human sensitive skin.

Biological systems are brilliant regarding their computational efficiency, complexity and adaptability. Therefore, we complete the system by carefully investigating biologically inspired cerebellum-based control strategies.

Join us! We are always looking for students (Praktikum, Master, Werkstudent, ...) to help us in our innovative research.