Biologically Inspired Reconfigurable Network of Networks

1
Biologically Inspired Reconfigurable Network of
Networks
N00014-99-1-0884 Jeffrey P. Sutton, M.D.,
Ph.D. MGH Neural Systems Group Harvard - MIT
Division of Health Sciences and
Technology sutton_at_nmr.mgh.harvard.edu Distr
ibuted and Network-Centric Autonomous
Collaborative Unmanned Vehicles Self-Sustaining
Intelligent Unattended Netted Force Summer
Review - August 1-3, 2001
2
Network of Networks 432 Networks

• Mapping of physical system onto system of dynamic
  neural networks
• High capacity connectivity
• Distributed, collaborative computation and data
  integration
• Compute within and between multiple scales
  simultaneously
• Adaptable, with multiple apps - sensory
  processing, global computation with local
  decision making, action and control in changing
  environment
• Reconfigurable networks for intelligent,
  autonomous operations

Sutton JP, Anderson JA. System and method for
high speed computing and feature recognition
capturing aspects of neocortical computation.
U.S. patent 5,842,190. 1998 Nov 24.
12 FNCs and Required Enabling Capabilities
3
AVSYS ModelAutonomous Vehicle SYStem for
coherent behavior among weakly interacting search
vehicles

• Version 1.0 - August 2000
• Version 2.0 - January 2001
• Version 3.0 - August 2001
• Grid Model - August 2001

AVSYS Version 2.0
4
The Grid Model

• GUI features of AVSYS Version 2.0
• AVs fly in a static formation and perform a grid
  search pattern
• There are no lead or wing AVs

5
Summary of Findings and Achievements

• Clear delineation of rules, inspired by
  neurobiological systems, that govern operations
  and communication in adaptable networks (i.e.,
  dynamic overlap functions)
• NoN applicable to sensor data processing and to
  reconfigurable, autonomous, multi-vehicle
  networking
• Comparison of consensus (Mode 1 AVSYS v 2.0)
  and aggressive (Mode 2 AVSYS v 3.0) network
  behavior
• Simulation of intelligent autonomy with minimal
  bandwidth / communication among AVs
• Development of a procedure for determining
  scalability
• Showed superiority (e.g., speed, fuel saving) of
  foraging / swarm behavior over grid searches in
  certain conditions
• Presentation at 5th ICCNS, and patent application
  filed