Evolutionary-Driven Agent Adaptation in Optimizing Search: Initial Results

Abstract

This research was an initial exploration into the ability of a team of autonomous software agents to be effective m unknown and changing optimization environments by evolving to use the most successful algorithm at the points in the optimization process where they will be the most effective. The goal of this project was to define the cure framework and methodology for agent-based adaptive optimization Strategies using an evolutionary approach at the strategic, rather than solution level - where the strategies of agents in the team (the decisions for picking, altering, and inserting a solution) evolve over time. As one application ot this approach, individual solutions are lours in the familiar combinatorial optimization problem of the traveling salesman. As another application, an initial 3 dimensional layout problem is defined The work also sets the stage to begin investigating how teams coordinate.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jun 30, 2009
Accession Number
ADA586720

Entities

People

  • Jonathan Cagan

Organizations

  • Carnegie Mellon University

Tags

Communities of Interest

  • Autonomy
  • Energy and Power Technologies
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Air Force
  • Algorithms
  • Application Software
  • Applied Computer Science
  • Artificial Intelligence
  • Autonomous Agents
  • Computations
  • Computer Science
  • Engineering
  • Environment
  • Genetic Algorithms
  • Mechanical Engineering
  • Multiagent Systems
  • Optimization
  • Software Agents
  • Students
  • Three Dimensional

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Operations Research
  • Systems Analysis and Design