CPS: Small: Generation of natural movement for a multiple degrees-of-freedom robot driven by stochastic cellular actuators
Lead PI:
Jun Ueda
Abstract
The objective of this research is to understand mechanisms for generating natural movements of skeletal mechanisms driven by stochastically-controlled, biologically-inspired actuators. The approach is to verify the hypothesis that the variability associated with high redundancy and the stochastic nature of the actuation is key to generating natural movements. This project seeks to: (i) develop a method to model and characterize actuator array topologies; (ii) develop a method to analyze the force variability of stochastic actuator arrays; (iii) develop an analytical method to generate movements for a robot with multiple degrees of freedom by minimizing the effect of variability; and (iv) demonstrate the validity of the approach through the development of a robotic arm driven by multiple stochastic array actuators. With respect to intellectual merit, the study of inhomogeneous stochastic actuator network topologies inspired by neuromuscular systems could find the "missing links" that bridge the gap between biological natural movements and the ones in artificial systems. Potential results could impact other research areas, including robust computer networks, robust immune systems, and redundant muscle coordination. With respect to broader impacts, a new graduate-level course provides students in engineering and science with a comprehensive and multidisciplinary education in the underlying principles, cutting-edge applications, and societal impacts of biologically-inspired robotics. Outreach activities include an interactive educational program for K-12 students and a workshop for high-school students and their mentors on robot development. International collaboration with Tokyo University of Science, Japan, will be initiated.
Performance Period: 09/01/2009 - 08/31/2013
Institution: GA Tech Research Corporation - GA Institute of Technology
Sponsor: National Science Foundation
Award Number: 0932208