This thesis is focused on the control and modeling of a robot finger. The control target of this project is position tracking during the finger motion. A new sliding mode control algorithm is developed in this project. It originates from the author’s 3-stages approach. i.e. the reaching phase, the sliding mode and the steady state. This approach is based on two key idea, as followings: (i) During the reaching phase, the speed of reaching can be related to the distance between the state trajectory and the switching manifold, (ii) During the sliding motion, the state velocity is directly related to the state variable for 2nd-order nonlinear systems. This new control algorithm is a functional relationship for the speed of reaching during the reaching phase. The simulation result demonstrates that this algorithm performs better than Gao and Hung’s “power rate” reaching law both in chattering reduction and reaching time. The above robot finger is developed via transition from a biological model for the human thumb to a mechanical prototype. A main concern of the biological model is the relationships on the excursion of finger tendons. The newly developed relationships in this project involve the shape and size of the associated phalanx, as well as the pulley mechanism besides the angle of rotation. The treatment of these relationships in this thesis are more complete and applicable than the approach by Landsmeer. Landsmeer’s model involves only the angle of rotation. Moreover, the mechanical prototype developed in this project retains anthropomorphic features of the human thumb. Such features include the quad-circles at the ends of each piece, and the very low friction in motion via artificial tendons. These properties are unique in the above robot finger compared to the work on robot hands by other research groups. In addition, SMA-wires are used as actuator for the above robot finger. A major problem of the SMA material is the “residue” stress. This is a defect since it prevents the wire from returning to the same position before re-activation in the next cycle. Hence a new approach: the “interval insertion” method has been developed in this project. Implementations show that this method really solves the above problem.
Author: Loh, Albert Ming
Source: City University of Hong Kong
Download URL 2: Visit Now