Simulation of a human finger joint with sliding surfaces

Abstract

In any automation process, object manipulation is a main task. The human hand and arm represent a system exceeding the versatility of current manipulation systems. Hence, at DLR a highly anthropomorphic hand arm system is being developed which aims at transferring the functionality of the respective human joints to a robotic system. The kinematics of human joints is often modelled with fixed rotational axes. This approximation leads to an error, which was found to be large especially for the CMC joint of the thumb. In this thesis, a new approach of simulating the kinematics of a human finger joint based on contacting surfaces is proposed. The CMC joint of the thumb is modelled as a multi body system in Simpack, consisting of one fixed and one unconstrained bone. The movement of the joint is determined by the forces exerted by the contacting articular surfaces, eight muscles and seven ligaments. The contact geometry is represented by undeformable polygonal surfaces extracted from MRI images. A modified version of PCM is used to compute the contact forces depending on the surface penetration depth. Muscles and ligaments are modelled as point-to-point nonlinear spring elements, exerting tensile forces along the connection line between a proximal and a distal attachment point. The attachment locations are determined based on anatomical and biomechanical literature. All components are assembled in Simpack to conduct simulations in time domain. The movement is controlled by eight input functions representing the activation levels of the muscles. Even though the present model is not optimized, comparison to literature shows that single and multi muscle activation leads to physiological movements. It is concluded that multi body simulations in combination with PCM represent a simple and fast means of simulating finger joint kinematics based its actual morphology. Due to its anatomical correspondence, the proposed model is capable of investigating the inuence of single tissues to joint movements, which makes it more versatile than conventional models based on fixed axes.