Docstoc

swarm_agalkin_final

Document Sample
swarm_agalkin_final Powered By Docstoc
					     Swarm Robotics
            Anton Galkin
    24779 Nano/Micro-Robotics
Department of Mechanical Engineering
     Carnegie Mellon University
       Pittsburgh, PA 15289
  Email: agalkin@andrew.cmu.edu
            Swarm Basics
• Motivation: Biomimicry
    Ants, birds, fish
• Decentralized local interactions
    no global information
• Behavior-based intelligence
    simple, inexpensive
    Advantages to Swarming (Nature)
•   Enhanced protection
•   Greater ease of travel
•   Predator confusion
•   Increased capability (perform tasks
    previously impossible or impractical)
    -carrying heavy objects
    -building structures many orders of
    magnitude greater than agent
Advantages to Swarming (Robotics)
• Redundancy & Failure tolerance
  -single agent failure is not catastrophic
• Decreased complexity (usually)
• Decreased cost (usually)
• Versatility, ease of adaptability
• Scalability
• Rapid wide-area coverage
• Increased capability
  -perform non-linear tasks
  -perform prohibitively expensive, complex or
   time consuming tasks more easily
               Biomimicry
• Inspiration:
  -social insects
  -schools of fish
  -flocks of birds
 Biomimicry - Pattern vs. Function
• Human perception can be misleading
• Evolutionarily neutral
  -funnel or torus swarm shapes
                - OR -
• Adaptive to group dynamics
  -coordinated movement & directed activity
Directed Activity
          Swarm Modeling
• Lagrangian method
• Swarm Aggregation
• Attractant-repellant model
  -autonomous agents modeled as inertial
   mass subject to forces from other agents
  -long range attraction
  -short-range repulsion
• Rule size or numerical preference
            Research methods
A typical scene
from a human
swarm day


“Using a
Collection of
Humans as an
Execution
Testbed for
Swarm
Algorithms”
       “Red Herring” Applet
• Java 2 SDK 1.4.2.05
  Swarm Modeling - Equations
• Attractant-repellant model
  -function of distance to considered agent
  -positive = attractant
  -negative = repellant
• Final choice: linear relationship


     atan(x-20)     sqrt(x-1)-4.5     x/2-10
 Swarm Modeling - Aggregation
                        1.2


                         1




Relative cluster size   0.8


                        0.6


         vs             0.4


                        0.2

    Population           0
                              1   3   5   7   9   11   13   15   17   19   21   23   25   27   29   31   33   35   37   39




                         5
                        4.5

                         4


Number of Clusters
                        3.5
                         3
                        2.5

       vs                2
                        1.5
                         1

   Population           0.5

                         0
                              1   3   5   7   9   11   13   15   17   19   21   23   25   27   29   31   33   35   37   39
 Static vs. Dynamic Equilibrium
• Static equilibrium
  -stable positions
  -no motion
  -geometrically optimal
  -(eqdist >> r)
• Dynamic equilibrium
  -constantly in motion
  -(eqdist > r)
            Predator Avoidance
• New “predator” agent
  -always repells                         -10/x

• inverse F-x relationship
• Interesting agent behavior




  herding       splitting      avoiding           vacuole
             Conclusions
• Simple aggregation models can lead to
  complex autonomous agent behavior
• Applied fish school dynamics?
  -localized interactions
  -minimalist intelligence/sensor array
  -inexpensive, disposable robots
• Collective swarm intelligence
                                           References
1.   Emma Alenius1, Åge J. Eide2, Jan Johansson1, Jimmy Johansson1, Johan Land1 and Thomas Lindblad1,
     “Experiments on Clustering using Swarm Intelligence and Collective Behavior” 1Royal Institute of Technology, S-
     10691 Stockholm, 2Ostfold College, N-1757 Halden,
2.   By Guy Theraulaz1, Jacques Gautrais1, Scott Camazine2 and Jean-Louis Deneubourg3, “The Formation of Spatial
     Patterns in Social Insects: From Simple Behaviors to Complex Structures,” 1CNR-FRE 2382, Centre de Recherches
     sur la Cognition Animale, Universite Paul Sabatier, 118 route de Narbonne, 31062 Toulouse Cedex 4, France;
     2Medical, Science and Nature Images, 310 West Main Street, Boalsburg, PA 16827-1327, USA; 3CENOLI, CP 231,
     Universite Libre de Bruxelles, Boulevard du Triomphe, 1050 Brussels, Belgium; 6 May 2003
3.   G. Dudek1, M. Jenkinj E. Milios2, and D. Wilkest3, “A Taxonomy for Swarm Robots,” 1Research Centre for Intelligent
     Machines, McGill University, Montrkal, Qukbec, Canada; 2Department of Computer Science, York University, North
     York, Ontario, Canada; 3Department of Computer Science, University of Toronto, Toronto, Ontario, Canada, 26 July
     1993
4.   C. Ronald Kube, Hong Zhang, “Collective Robotic Intelligence,” Department of Computing Science, University of
     Alberta, Edmonton, Alberta Canada T6G 2J9, 1 Sept 1992
5.   Debashish Chowdhury1, Katsuhiro Nishinari2, and Andreas Schadschneider3, “Self-organized patterns and traffic
     flow in colonies of organisms: from bacteria and social insects to vertebrates,” 1Department of Physics, Indian
     Institute of Technology, Kanpur 208016, India; 2Department of Applied Mathematics and Informatics, Ryukoku
     University, Shiga 520-2194, Japan; 3Institute for Theoretical Physics, Universit¨at zu K¨oln, 50937 K¨oln, Germany, 9
     January 2004
6.   Erol Sahin, “Swarm Robotics: From Sources of Inspiration to Domains of Application,” KOVAN – Dept. of Computer
     Eng., Middle East Technical University, Ankara, 06531, Turkey, erol@ceng.metu.edu.tr, E. Sahin and W.M. Spears
     (Eds.): Swarm Robotics WS 2004, LNCS 3342, pp. 10–20, 2005.
7.   Julia K Parrish1,2, Steven V. Viscido2, Daniel Gru Nbaum3, “Self-Organized Fish Schools: An Examination of
     Emergent Properties,” 1School of Aquatic and Fishery Sciences, Box 355020, University of Washington, Seattle,
     Washington, 98195-5020; 2Zoology Department, University of Washington; and 3School of Oceanography,
     University of Washington, Biol. Bull. 202: 296–305., June 2002
8.   Y. LIU, K. M. PASSINO, Communicated by M. A. Simaan, “Biomimicry of Social Foraging Bacteria for Distributed
     Optimization: Models, Principles, and Emergent Behaviors,” Journal of Optimization Theory and Applications: Vol.
     115, No. 3, pp. 603–628, December 2002
9.   Daniel W. Palmer, Mark Kirschenbaum, Jon P. Murton, Michael A. Kovacina*, Daniel H. Steinberg**, Sam N. Calabrese, Kelly M. Zajac,
     Chad M. Hantak, Jason E. Scatz, “Using a Collection of Humans as an Execution Testbed for Swarm Algorithms,” John Carroll University,
     University heights, OH 44118, *Orbital Research Inc, Highland Heights, OH 44143, **Dim Sum Thinking Inc, University Heights, OH 44118,

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:3
posted:10/28/2011
language:English
pages:16