Lecture 1 - Introduction
What is an Agent?
● An agent is a computer system that is situated in some environment, and that is capable
of autonomous action in this environment in order to meet its delegated objectives
● Note: autonomy is a spectrum!
Multi-Agent Systems, a Definition
● A Multi-Agent System is one that consists of a number of agents that interact (with each
other and the environment)
● In general, agents will have different goals (often conflicting!)
● To successfully interact, they will have to learn, cooperate, coordinate, and negotiate
Agents and Environment
Motivations for studying MAS
● Techological:
○ Growth of distributed, networked computer systems
■ (computers act more as individuals than parts)
○ Robustness: no single point of failure
○ Scalable and flexible:
■ adding new agents when needed
■ asynchronous, parallel processing
○ Development and reusability
■ components developed independently (by specialists)
● Scientific:
○ Models for interactivity in (human) societies,
■ e.g. economics, social sciences
○ Models for emergence of cooperation
1
, ■ Coordination: cooperation among non-antagonistic agents
■ Negotiation: coordination among self-interested agents
Application: Robotics
● Robots as Physical Agents (Embodiment)
○ Internet of Things (IoT)
○ Swarms of drones,
○ Fleet of autonomous vehicles
○ Physical internet
Multiagent Systems: typical scientific questions addressed
● How can cooperation emerge in societies of self-interested agents?
● What actions should agents take to optimize their rewards/utility?
● How can self-interested agents learn from interaction with the environment and other
agents to further their goals?
● How can autonomous agents coordinate their activities so as to cooperatively achieve
goals?
MAS as Distributed AI (DAI)
● AI : Cognitive processes in individuals
○ Inspiration: neuro-science, behaviourism, ...
● DAI: Social processes in groups
○ Inspiration: social sciences, economics, ....
● Basic question in DAI
○ How and when should which agents interact (compete or collaborate) in order to
achieve their design objectives?
● Approaches:
○ Bottom-up: given specific capabilities of individual agents, what collective
behaviour will emerge?
○ Top-down: Search for specific group-level rues (e.g., conventions, norms, etc.)
that successfully constrain or guide behaviours at individual level;
Multiagent Systems is Interdisciplinary
● The field of Multi-Agent Systems is influenced and inspired by many other fields:
○ Economics
○ Game Theory
○ Philosophy and Logic
○ Mathematics (e.g. optimal control)
○ Ecology
○ Social Sciences
● This can be both a strength and a weakness
● This has analogies with Artificial Intelligence itself
2
, Intelligent Agents
● An intelligent agent is a computer system capable of flexible autonomous action in some
environment
● Autonomous: not pre-determined by designer
● By flexible, we mean:
1. Reactive (able to receive information from environment and respond)
2. Pro-active able to reason and/or learn and work towards goals)
3. Social (able to communicate, coordinate, negotiate and cooperate)
Simple Typology for Intelligent Agents
● Intelligence in agents covers a spectrum:
● Reflex agents
○ Simple reflex agents
○ Model-based reflex agents
● Goal based agents
● Utility based agents
● Learning agents
Type 1: Simple Reflex Agent
● Reacts to environment
○ Percept → Action
○ Based on simple if-then rules
(condition-action)
● Properties:
○ No state: ignore history
○ Pre-computed rules
○ NO Partial observability
Type 2: Model-Based Reflex Agent
● Reflex agent with state
● Agent uses memory to store an internal representation of its world
● Internal model based percept history
● This internal model allows him to handle partially observable environment
Type 3: Goal-Based Agent
● Goal = desired outcome
● Goal-based (planning) agents act by reasoning about which actions to achieve the goal
● Less efficient, but more adaptive and flexible
● Search and planning: AI subfields concerned with finding sequences of actions to reach
goal.
3
What is an Agent?
● An agent is a computer system that is situated in some environment, and that is capable
of autonomous action in this environment in order to meet its delegated objectives
● Note: autonomy is a spectrum!
Multi-Agent Systems, a Definition
● A Multi-Agent System is one that consists of a number of agents that interact (with each
other and the environment)
● In general, agents will have different goals (often conflicting!)
● To successfully interact, they will have to learn, cooperate, coordinate, and negotiate
Agents and Environment
Motivations for studying MAS
● Techological:
○ Growth of distributed, networked computer systems
■ (computers act more as individuals than parts)
○ Robustness: no single point of failure
○ Scalable and flexible:
■ adding new agents when needed
■ asynchronous, parallel processing
○ Development and reusability
■ components developed independently (by specialists)
● Scientific:
○ Models for interactivity in (human) societies,
■ e.g. economics, social sciences
○ Models for emergence of cooperation
1
, ■ Coordination: cooperation among non-antagonistic agents
■ Negotiation: coordination among self-interested agents
Application: Robotics
● Robots as Physical Agents (Embodiment)
○ Internet of Things (IoT)
○ Swarms of drones,
○ Fleet of autonomous vehicles
○ Physical internet
Multiagent Systems: typical scientific questions addressed
● How can cooperation emerge in societies of self-interested agents?
● What actions should agents take to optimize their rewards/utility?
● How can self-interested agents learn from interaction with the environment and other
agents to further their goals?
● How can autonomous agents coordinate their activities so as to cooperatively achieve
goals?
MAS as Distributed AI (DAI)
● AI : Cognitive processes in individuals
○ Inspiration: neuro-science, behaviourism, ...
● DAI: Social processes in groups
○ Inspiration: social sciences, economics, ....
● Basic question in DAI
○ How and when should which agents interact (compete or collaborate) in order to
achieve their design objectives?
● Approaches:
○ Bottom-up: given specific capabilities of individual agents, what collective
behaviour will emerge?
○ Top-down: Search for specific group-level rues (e.g., conventions, norms, etc.)
that successfully constrain or guide behaviours at individual level;
Multiagent Systems is Interdisciplinary
● The field of Multi-Agent Systems is influenced and inspired by many other fields:
○ Economics
○ Game Theory
○ Philosophy and Logic
○ Mathematics (e.g. optimal control)
○ Ecology
○ Social Sciences
● This can be both a strength and a weakness
● This has analogies with Artificial Intelligence itself
2
, Intelligent Agents
● An intelligent agent is a computer system capable of flexible autonomous action in some
environment
● Autonomous: not pre-determined by designer
● By flexible, we mean:
1. Reactive (able to receive information from environment and respond)
2. Pro-active able to reason and/or learn and work towards goals)
3. Social (able to communicate, coordinate, negotiate and cooperate)
Simple Typology for Intelligent Agents
● Intelligence in agents covers a spectrum:
● Reflex agents
○ Simple reflex agents
○ Model-based reflex agents
● Goal based agents
● Utility based agents
● Learning agents
Type 1: Simple Reflex Agent
● Reacts to environment
○ Percept → Action
○ Based on simple if-then rules
(condition-action)
● Properties:
○ No state: ignore history
○ Pre-computed rules
○ NO Partial observability
Type 2: Model-Based Reflex Agent
● Reflex agent with state
● Agent uses memory to store an internal representation of its world
● Internal model based percept history
● This internal model allows him to handle partially observable environment
Type 3: Goal-Based Agent
● Goal = desired outcome
● Goal-based (planning) agents act by reasoning about which actions to achieve the goal
● Less efficient, but more adaptive and flexible
● Search and planning: AI subfields concerned with finding sequences of actions to reach
goal.
3
