Implementation and Analysis of Neuromorphic Motor-Pattern Generating Systems

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

The generation of rhythmic movement is essential to locomotion and many other activities in most animals. The pattern-generating networks in the brain that control these movements are efficient and reliable, and so provide a platform to study a critical set of biological control paradigms, and offer the potential to inspire engineered systems that exploit these underlying principles. Neuromorphic engineering is the development of hardware-based systems that are inspired by circuit architectures found in biological nervous systems. The goals of this collaborative project are to develop neuromorphic systems that emulate the control and production of such rhythmic movements, and to use nonlinear dynamical analysis to gain a better understanding of the production of these movements. Building on prior development of very large-scale integrated (VLSI) circuits that mimic motor patterns, and channel-based model neurons using such circuits, this project will create neuromorphic oscillators based on the dynamics of ionic channels in biological neurons, combine these oscillators with mechanical actuation and sensory feedback, and expand these systems to include heterogeneous populations of neurons. The models will test the engineering validity of some principles inferred from biological designs, and will in turn allow quantitative testing of how varied parameters, sensory feedback, and heterogeneous model populations may have effects in actual complex biological systems. Results from this work will extend beyond neuroengineering, to a better understanding of motor control and feedback in neurobiology, and to potential applications in robotic design. This collaborative project also continues training of students from various backgrounds and at various levels in laboratories with a solid history of interdisciplinary training.