Introduction to AI - Week 10

Reactive Systems - Rational Agents


Key Concepts

Inspired by biological systems:

Principles of Computation


Reactivity - Example

Herbert - a robot to find soda can-like objects:

Example (Cont'd)


Open problems:

Compromise: Reaction-First Search

Assume situated agent: system consists of a planner and a reactor.

The Problem

The Principles of Reaction First Search

That is, the planner knows the policy and explores the search space searching in the policy first.

Then: Performance of rfs is monotonically increasing.

Example Scenario

Task: phase A - start at I go to G, grasp something, and phase B - go back to I.

Policy: in phase A and B reduce Manhattan distance between current position and G and I respectively.

Note: non-deterministic behaviour, in particular the reactor might be trapped.

The Planner in the Example


Intelligent Agents

The concept "agent" is very wide and its usage varies. It may mean:

Depending on the level of sophistication an agent should have, we need different architectures. We will look at the architecture of a learning agent, a more sophisticated single-agent architecture, and a multi-agent scenario.

Learning Agent

Russell-Norvig: An Agent is anything that perceives its environment through sensors and acts upon that environment through effectors.

Learning agent has learning element and performance element

Russell/Norvig's Learning Agent

A General Agent Architecture by A. Sloman

Characteristics of Intelligent Agents

[Shaun Abushar, Naoki Hirata]

Agents May Coordinate their Actions

Assume a set of agents, which make plans. and a centralised coordinator which negotiates with agents about how to coordinate their distributed plans.


Example: Assume a transportation task of delivering 10,000 flowers from New Jersey to New York City. The task may be performed by several trucks or by a single helicopter, a lift-truck is necessary in both cases. A coalition of trucks and lift-trucks will probably be cheaper than a coalition of a helicopter and lift-trucks.

It is possible to automatically allocate the tasks by the formation of coalitions.

A coalition is a group of agents that have decided to cooperate in order to achieve a common task.

Planning in Coalitions

Why Mobile Agents?

[Danny B. Lange and Mitsuru Oshima 1999]

Why Mobile Agents? (Cont'd)


Further Reading    

There is a lot of material on-line, e.g.

© Manfred Kerber, 2004, Introduction to AI
The URL of this page is
URL of module