Reduction of dimensionality of a cellular actuator array for driving a robotic hand
Author(s)Cho, Kyu-Jin, Ph. D. Massachusetts Institute of Technology
Massachusetts Institute of Technology. Dept. of Mechanical Engineering.
H. Harry Asada.
MetadataShow full item record
In an attempt to explore an alternative to today's robot actuators, a new approach to artificial muscle actuator design and control is presented. The objective of this research is to coordinate the multitude of artificial muscle actuator axes for a large DOF (degree of freedom) robotic system based on dimensionality reduction. An array of SMA actuators is segmented into many independently controlled, spatially discrete volumes, each contributing a small displacement to create a large motion. Segmented Binary Control is proposed where each segment is controlled in an on-off manner, creating a stepper-motor like actuator. This overcomes hysteresis and other nonlinearities of the actuator material. The segmented cellular architecture of SMA wires is extended to a multi-axis actuator array by arranging the segments in a two-dimensional array. The multi-axis control is streamlined and coordinated using a grouping of segments called C-segments in order to activate multiple links of a robot mechanism in a coordinated manner. This allows control of large DOF with a small number of controls. The proposed approach is inspired by the segmented architecture of biological muscles and synergies, a strategy of grouping output variables to simplify the control of large number of muscles. Data from various hand postures are collected using data glove and used in creating the C-segment design that is capable of performing the given postures. A lightweight Robotic Hand with 16 DOF is built using shape memory alloy actuators. This hand weighs less than 1kg including 32 SMA actuators and control circuitry. Eight C-segments that are ON-off controlled are used to create sixteen given postures. In the future, this approach can be applied to applications where the control signal is inherently limited due to limited amount of information that can be extracted or transferred to the robot, such as brain machine interface and tele-operation.
Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Mechanical Engineering, 2007.Includes bibliographical references (p. 89-93).
DepartmentMassachusetts Institute of Technology. Department of Mechanical Engineering
Massachusetts Institute of Technology