August 19-20, 2010

Center for Nonlinear Studies
TA-03, Building 1690
Los Alamos National Lab

 

Abstract:

Many modern systems contain multiple goal-directed agents. Often some of the agents in these systems know the goals of other agents, and use this knowledge to guide their behavior, which makes their design and control particularly challenging. Some examples of systems of goal-directed agents include:

• A smart power grid, where the agents are power producers, power distributers, grid operators, market managers, and power consumers, both corporate and individual.

• The national airspace, where the agents are airlines, traffic controllers, pilots, and travelers.

• A financial system, where the agents are firms, banks, trading funds, and private citizens.

• A battlefield, where the agents are combatants and commanders, together with autonomous, artificial military systems;

• Multi-disciplinary design and optimization systems, where the agents are the different design teams.

• Human-mechanical hybrid systems using robots in a coordinated disaster response, where the agents are the human operators together with the robots.

• Cyber environments designed to maximize productivity while limiting general or specific cyber attacks, where the agents are both human and artificial.

This workshop will focus on how to predict, design and control such systems of goal-directed agents. This is a wide-open area, situated at the intersection of game theory, optimization (both closed form and computational), reinforcement learning, Bayes nets, AI more generally, and many other fields.

This workshop is being held in conjunction with a more general 3-day workshop at the Santa Fe Institute that focuses on more general concepts related to interacting goal-directed agents. We are also hosting a following week long working-group activity during August 23-27, 2010 at the Center for Nonlinear Studies to explore specific problem solving and collaboration building opportunities. Participation is by invitation only for external participants.