Neuromorphic Motor-Pattern Generating Systems
Primary Investigator:
Dr. Stephen P. DeWeerth
Other Group Members
:Mike Reid, Michael Sorensen, Shane Migliore, Edgar Brown
Collaborators
:Rob Butera, Ron Calabrese, Bill Ditto
Biological systems present powerful paradigms for controlling movement. Even very simple animals are able to navigate through complex and changing environments, producing energy-efficient movements. One particularly important class of stereotyped movements, rhythmic movement, is used to produce effective locomotion (and other actions) in organisms ranging from mollusks to mammals. The pattern-generating networks that control these movements have been optimized to be both efficient and reliable; thus, they provide a platform for studying a set of biological control paradigms, and offer the potential for inspiring engineered systems that exploit these underlying principles.
The long-term goals of our research are to extract general biological control paradigms and to apply these principles to the design of systems that perform complex coordinated movements. Central to this pursuit is the development of neuromorphic systems: hardware-based systems that take their inspiration from circuit architectures found in biological nervous systems which impose physical constraints like those in their biological counterparts. These systems provide the following: (1) tools for developing a better understanding of the relationship between architecture and behavior in biological organisms, (2) platforms for developing real-time, complex systems for addressing engineering applications, and (3) biological information that is not easily attainable from the real organism.
Neuromorphic analog very large-scale integrated (aVLSI) circuits are used as building blocks for creating these neurally-inspired motor control systems. A simplified model neuron (Figure 1) is based on a single-compartment Hodgkin-Huxley (HH) formalism, and its dynamics are described completely by an autonomous set of differential equations.
Figure 1
A neuromorphic half-center oscillator (Figure 2) is built from these neurons, and the system is analyzed using nonlinear-dynamical techniques.
Figure 2
The neuromorphic oscillator will control a mechanical system (Figure 3), and the effects of sensorimotor feedback on the dynamical behavior of the system is analyzed.
Figure 3
Additionally, a network in which each neuron in the half-center oscillator is replaced by a population of neurons (Figure 4) is analyzed to study the effects of heterogeneity on the robust production of movements.
Figure 4
Acknowledgements
This work is being supported by the National Science Foundation (NSF Proposal Number 0131612).