The Laboratory for Intelligent Agents has two research focuses.
First, it aims to develop team-based agent
technologies that facilitate
a distributed team to collaborate and make decisions in a dynamic environment
by proactively exchanging and fusing information based on a "shared mental model"
about the team. The second focus of our lab is to develop market-based agent
technologies for addressing critical issues regarding negotiation,
resource allocation, supply chain management, and web services.
Our lab's research on team-based agents has built on the findings and
theories about effective human team behaviors. Psychologists who study
human teamwork have pointed out that intelligent team behaviors rely on
overlapping shared mental models among team members. For example,
a competent human team member can anticipate the needs of teammates,
offer relevant information proactively, and avoid overloading teammates by helping them.
Therefore, a key focus of the Laboratory for Intelligent Agents is
to develop technologies that enable agents in a team to exhibit intelligent
team behavior using a computational shared mental model.
To launch our research we have developed a team-based agent architecture
(Collaborative Agents for Simulating Teamwork).
agents anticipate information needs of teammates based on their
knowledge about the structure and process of the team, as well as their
beliefs about their teammates and the environment.
Our lab's research on market-based agents provides an analytic framework
for developing systems that can a) interact successfully with other agents
and humans in an e-commerce marketplace, b) use market information
to dynamically allocate resources to those who can use them most productively,
or c) use "prediction" markets (such as futures markets) in decision-making
that can aggregate participants information so that the market system knows,
in some sense, more than any single participant does.
A market-savvy agent will be able to interact with both humans and
electronic marketplaces to acquire the necessary resources to accomplish
its tasks in the most efficient and cost-effective manner possible.
R-CAST agents uses Klein's RPD cognitive model to capture the process
of decision making team, which include anticipating, seeking, and facing
information related to decisions.
Current research projects of the Laboratory for Intelligent Agents
investigate formal foundations, technologies, and applications of
team-based agents and market-based agents. These projects are currently
supported by the Air Force Office of Scientific Research (MURI),
National Science Foundation (ITR), the Army Research Office, and
the Army Research Lab. Previous sponsors include DARPA and USMC/PTC.
Director, Dr. John Yen
Associate Director, Dr. Tracy Mullen