research topics

Multiagent Systems

Multiagent systems arise in human societies, biological ecosystems, the immune system, teams of robots and distributed computation. While different in detail, these systems all produce complex global behavior through the local interactions of their constituent parts, each of which has only a limited view of the system as a whole.

Cooperation in Societies

• social dilemmas: encouraging cooperation among self-interested agents
• a model for the evolution of informal organizations (communities of practice) with a mechanism for adaptation to changes based on dynamical instability
  B. A. Huberman and T. Hogg, Communities of Practice: Performance and Evolution, 1995
• privacy-protecting mechanisms can encourage cooperation by reducing its cost to individuals

Computational Societies and Economies

Large collections of computational agents have many analogies to market economies and biological ecosystems in that the individuals make decisions based on local information that is often incomplete, out of date and uncertain.

• dynamics and control of multiagent systems
• cooperative problem solving
• Market-like mechanisms can form the basis for flexible distributed software systems and for allocation of computational resources:
  C. A. Waldspurger et al., Spawn: A Distributed Computational Economy, 1992

Agent-Based Control of Smart Matter

Smart matter consists of sensors, actuators and computers embedded in physical materials. These devices can alter physical properties of the material under program control. When the number of devices is large, multiagent systems can provide robust distributed control through analogies with biological ecosystems, economic markets and the scientific community.

• control of smart matter