UIUC AGENTS AND MULTI-AGENT
SYSTEMS GROUP (AMAG)
The people of the UIUC Agents and Multi-Agents Group
(AMAG) are investigating theories, methods, technologies, and applications
of actor-based, agent-oriented, and multi-agent information systems.
Current participants: Professors Gul
Agha (CS), Les Gasser (GSLIS), Mike Shaw (Commerce/Business); John Barron,
Nadeem Jamali, Eric Rankin, Gek-Woo Tan, Prasanna Thati, Carlos Varela,
James Waldby
DEFINITIONS:
"Agents" are generally computational
objects with the following characteristics:
-
Autonomy: a local thread of control
-
Structured, persistent action
-
Sophisticated, resource-controlled reasoning and rationalized
choice
-
Integrated, multifunction architectures
-
Acting in complex, dynamic domains and environments
(e.g., dynamics, mobility, openness)
-
Adaptive, learning capabilities
"Social Agents" work toward the ability to interact
with other agents and with people, exhibiting characteristics such as:
-
Emotion
-
Personality
-
Believability
-
Multimodal interaction
"Multi-Agent Systems" are agent collectives,
possibly involving people as participants, whose key
problematics are:
-
Dynamic division of labor: who does what, when
-
Coordination of activity:
-
Management of interdependencies
-
Teams, groups, organizations
-
Managing heterogeneity
CURRENT RESEARCH TOPICS IN MAS:
Interaction capabilities, constraints, and prefererences
:
-
Communication languages and protocols
-
Organization and social structure
-
Teamwork and cooperation
-
Decentralized systems
-
Mechanism design
Reasoning about coordinated interactions
-
Conflict resolution and negotiation
-
Multi-agent planning
-
Coalition formation and organization self-design
-
Agent modeling and plan recognition
-
Multi-agent learning
-
Distributed search and constraint satisfaction
-
Foundations (multi-agent logics, game-theory, economics,
philosophy, etc.)
Engineering, deploying, and evaluating multi-agent systems
-
Agent programming languages
-
Multi-agent programming frameworks
-
Agent models and architectures
-
Standards for multi-agent technology (interaction protocols,
languages)
-
Development and engineering methodologies
-
Evaluation of multi-agent systems
-
Testbeds and development environments
-
RoboCup, Multi-agent robotics, RoboCup-Rescue
-
User interfaces and personalizable agents
Practical applications
-
Agents in electronic commerce
-
Information gathering and management
-
Distributed resource allocation
-
Multi-agent simulations of social and biological systems
-
Multi-agent vision and robotics
-
Interacting personal digital assistants
CURRENT INTERESTS/PROJECTS OF UIUC-AMAG MEMBERS
Gul Agha (CS): Programming models for
parallel, distributed and mobile computing.
Les Gasser (GSLIS) Social knowledge,
action, and learning; language evolution; MAS research tools and methods;
networked information systems and information management; computational
organization research.
Mike Shaw (Commerce): Multi-agent frameworks
for enterprise information modeling and supply-chain analysis; SWARM (SFI)
models of organizations and supply-chain networks as complext adaptive
systems; genetic algorithm models of group decision and learning (with
Prof. Riyaz Sikora).
John Barron: Evaluating business-to-business
electronic commerce
Nadeem Jamali: Agent architectures
for supporting market-based mechanisms for controlling resource consumption
by mobile agent ensembles.
Eric Rankin: Language evolution:
how differing agents can generate a shared lexicon and learn to
recognize emerging communication patterns (language).
Gek-Woo Tan: Modeling supply-chains
as complex adaptive systems
Prasanna Thati:
resource management and security in mobile agent systems
Carlos A. Varela: A hierarchical model
of coordination of concurrent activities and its application in the distributed
context of worldwide computing.
James Waldby: Information security
issues in agent and actor software