Kinematic Modelling of the Human Hand for Robotics

Abstract

Human hand movement models are needed for the design of humanoid robotic hands. For the first time, a complete human hand movement model is created on the basis of magnetic resonance imaging (MRI). The model is applied in a simulation of grasps. MRI and optical motion capture of surface markers (MoCap) are compared to find out which is better suited for kinematic hand modelling. They perform similarly well. However, a skin movement model can be used to further reduce the MoCap errors near joints by about 50%.