Multiagent
Systems:
Local Decisions
vs.
Global
Coherence
Leen-Kiat Soh, Nobel Khandaker, Adam Eck
Computer Science & Engineering
University of Nebraska
Lincoln, NE 68588-0115
lksoh@cse.unl.edu
http://www.cse.unl.edu/agents
November 25, 2008
Agents
• What is an agent?
– An agent is an entity that takes sensory input from
its environment, makes autonomous decisions,
and carries out actions that affect the environment
– A thermostat is an agent
– A calculator is not an agent
Agent
sensory
input
think!
Environment
output
actions
Intelligent Agents
• What is an intelligent agent?
– An intelligent agent is one that is capable of flexible
autonomous actions in order to meet its design
objectives, where flexibility means:
• Reactivity: agents are able to perceive their environment, and
respond in a timely fashion to changes that occur in order to
satisfy their design objectives
• Pro-activeness: agents are able to exhibit goal-directed
behavior by taking the initiative in order to satisfy their design
objectives
• Social ability: agents are capable of interacting with other
agents (and possibly humans) in order to satisfy their design
objectives
(Wooldridge and Jennings 1995)
Intelligent Agents: Learning
• Machine Learning in AI says
The acquisition of new knowledge and motor and cognitive skills
and the incorporation of the acquired knowledge and skills in
future system activities, provided that this acquisition and
incorporation is conducted by the system itself and leads to an
improvement in its performance.
• Agents that learn are intelligent
• Not all agents are intelligent!
Agent Environment
• Inaccessible vs. accessible
– Incomplete vs. complete data
• Deterministic vs. non-deterministic
– Certainty vs. uncertainty
• Episodic vs. non-episodic
– Each episode is independent or not
• Static vs. dynamic
– Remain unchanged except by the performance
of actions by the agent?
• Discrete vs. continuous
– “Chess game” vs. “taxi driving”
Why Agents?
• If the system-to-be-built has, during the
execution of the system
Why does a person hire an agent?
– Incomplete data
– Uncertainty in the assessment/interaction of its
environment
– Inter-dependent episodes of events
– No full control over the events in the environment
– An “open world”, instead of a “closed world”
• In other words, agents are used when you need
to build a system that is adaptive to an uncertain,
dynamic, and at times unexpected environment
– So you can make full use of the autonomous property
of an agent
Multiagent Systems
• A multiagent system is a system where multiple
agents perform a task better when working
together
– Interaction (communication)
– Coordination
– Collaboration
• Example: A group of basketball players who do
not observe or communicate with each other is
not a team—simply a group of individual agents.
Why Multiagent Systems?
• Distributed control, databases, knowledge bases,
experience, expertise, execution, planning, and
views of environment
– Scalability, flexibility, autonomy
From the Solution point of view
• Distributed control, proprietary concerns, security
levels, communities
– Efficiency, flexibility, autonomy
From the Logistics point of view
Local Decisions vs. Global Coherence
• Agents making local decisions have high
autonomy
– Less reliant on other agents
– No explicit global control
– May lead to unexpected, “chaotic” results due to lack
of coordination
• A multiagent system should strive for global
coherence
– How well a system behave as a unit
Local Decisions vs. Global Coherence
High Coherence
DESIRABLE
NOT DESIRABLE
Low Coherence
Low Autonomy
High Autonomy
Multiagent Simulation Example - I
• Project – Multiagent System for Mining the
Asteroids
– Simulated 2D space with asteroids containing ores
– Scout agents search and Miner agents extract deposit
• Local Decision
– Limited Time and resource (scouts, miners, etc.)
– How to explore the space?
– How to assign miners to asteroids?
• Global Coherence
– Maximize average value per cargo unit
Multiagent Simulation Example - II
• Project – Multiagent System Simulating Immune
System of an Organism
– Pathogens eat Red Blood Cells and Multiply, White
Blood Cells Kill Pathogens
– System health depends on number of Pathogens
• Local Decision
– Pathogens: eat the nearby red blood cells
– White Blood Cells: Destroy the nearby Pathogen and
Call for Help
• Global Coherence
– System Health
Multiagent Simulation Example - III
• Project – Multiagent System Simulating an
Inferno
– 2D Space with Flammable Objects and Firefighters
• Local Decision
– Fire Burns the Nearby Objects
– Firefighters Put Out Fire and Help Other Firefighters
• Global Coherence
– Put Out All Fire
– Maximize the Health of the Firefighters
Multiagent Simulation Example - IV
• Project – Multiagent System Simulating
Adversarial Search
– 2D Maze Containing Diplomats, Terrorists, and Allies
– Terrorist and Allies Search a Maze and Try to Capture
Diplomats
• Local Decision
– Terrorists Share a Partial Map of the Maze
– Allies Share Whole Map of the Maze
• Global Coherence
– The Number of Diplomats Rescued By Allies
Multiagent Simulation Example - V
• Project – Collaborative Behavior in an Online
Learning Environment
– One teacher who assigns A collaborative learning activities
– S students who can learn from one another or by themselves
during an activity, and
– S student agents, each assigned exclusively to a particular
student to make decisions to improve their student’s learning
• Local Decision
– Choose learning activities for students
– Voting for permanent groups between students
• Global Coherence
– Average student knowledge converges to high levels
– Long lasting student learning groups
Local Decisions vs. Global Coherence
High Coherence
INFERNO
PATHOGENS
ASTEROIDS
TERRORISTS
ALLIES
STUDENT AGENTS
DESIRABLE
NOT DESIRABLE
Low Coherence
Low Autonomy
High Autonomy