Dynamic Modelling of a modular robot finger

Kurzfassung

New generations of dexterous robotic hands with high degrees of freedom, integrated advanced sensors, improving power-to-weight ratio, have shown great potential for performing complex tasks beyond simple grasping such as manipulation and gaiting. With the help of control strategies such as impedance and compliance control, together with various compensation methods, researchers have achieved some success in performing dexterous tasks with robotic hands. However, the lack of a dynamic model of the robotic hand has placed an inherent limitation on their dexterous performance. The goal of this work is to address this omission to help eliminate this control performance limitation by deducing a dynamic model for one of the more successful and established dexterous robotic hands available today: the DLR/HIT Hand II. This robotic hand consists of five identical modular fingers, each finger has three degrees of freedom with dedicated position sensors and force-torque sensors at every joint. As a result of its identical modularity, it is sufficient to model one finger to be extended to five fingers to obtain the model of the hand. Model based control is used in many applications in robotics, including the closely related robotic arms. However, for anthropomorphic robot hands little research is available. The contribution of this work is to create a complete model including a characterization of the motors to find the relationship between the control signal and the produced torque, the effects of friction in all the components, and the influence of the dynamics such as mass matrix, centrifugal term and gravitation. Furthermore, as the friction differs from finger to finger and the parameters may change over time, an algorithm to adjust the friction model is proposed in this work.