Evolvierende Schwärme zur Bildung von 3D-Strukturen (Evolving Swarms that Build 3D Structures)

Student:Sebastian von Mammen
Title:Evolvierende Schwärme zur Bildung von 3D-Strukturen (Evolving Swarms that Build 3D Structures)
Type:study thesis
Advisors:Kókai, G.
State:submitted on February 3, 2005
Prerequisits:
Topic:

Background:
Swarm intelligence systems are attracting more and more attention recently, as many design decisions nowadays have to incorporate dynamical systems with a very large number of interacting agents. One example is the design of city and, in particular, traffic infrastructures. Questions of where to build new roads, bridges, tunnels, intersections, etc is largely dependant on simulations of traffic patterns, as obtained through the monitoring of car and pedestrian traffic. Simulations, in which emergent traffic patterns are analyzed, have to be performed to evaluate different options for improved or additional infrastructure components. Hence, this iterative (and in parts evolutionary) design process involves collective intelligence systems (such as car agents or pedestrian agents). The Evolutionary & Swarm Design group at the Department of Computer Science, University of Calgary, is investigating simulations of emergent patterns in swarm intelligence systems. We are especially interested in models of social insects (army ants, termites), bacterial ecosystems, and of biomolecular systems (such as protein-protein interactions within a cell).

Thesis Objective
For the context of this student thesis we will focus on simulations of social insect behaviours. In particular, we want to investigate building behaviours of social insects, such as the construction of termites' nests. In [1] and [2] an evolutionary system is described, which programs swarm behaviours through interactive 'breeding'. The implemented system uses an evolutionary kernel (Evolvica [3, 4]) and an agent-based swarm simulator (Breve [5]). The thesis objective is to build on these two systems to implement an evolutionary design system for swarms, that model social animals (such as insects or birds) and their building behaviours (such as termites building their nests). The approach should, however, not focus on how a particular swarm system performs its building tasks, but rather explore an evolutionary way for getting the swarms to construct 3-dimensional structures. This could, for example, be achieved by having the swarm agents move small, stackable objects in a 3D environment. The actual control programs will be not manually designed but by interactive and/or automized breeding, following the well-tested approach in [2].

Literatur:

  • Kwong, H. and C. Jacob: Evolutionary Exploration of Dynamic Swarm Behaviour In Congress on Evolutionary Computation, 2003, Canberra, Australia: IEEE Press.
  • Kwong, H: Evolutionary Design of Implicit Surfaces and Swarm Dynamics Master Thesis, University of Calgary, May 2003.
  • Jacob, C: Illustrating Evolutionary Computation with Mathematica Morgan Kaufmann Publishers, San Francisco, 2001.
  • IEC web site: http://pages.cpsc.ucalgary.ca/~jacob/IEC
  • Jon Klein http://www.spiderland.org/breve
watermark seal