Semester 2 G54DIA Designing Intelligent Agents

This module gives a basic introduction to the analysis and design of intelligent agents, software systems which perceive their environment and act in that environment in pursuit of their goals.

Structure of the Module

The module consists of two parts

• lectures: covering the design of intelligent agents; and
• project work: involving the specification, design and implementation of one or more simple agents.

Lectures

Lectures are on Mondays at 16:00 and Tuesdays at 14:00 in room C60 (CS).

In week two only (week commencing 6th of February), both lectures will be on Monday: the first lecture will be at 14:00 in the tutorial slot, and the second lecture will be at 16:00 in the usual lecture slot.

Project Work

Each student must specify, design and implement one or more simple agents as described in the Project Description lecture.

A Java agent package (Package uk.ac.nott.cs.g54dia) is provided as a starting point for your project work.

Tutorials

The project work is supported by group and individual tutorials. Tutorials start in week 4 (week beginning the 20th of February). The group tutorials cover the use of the Java agent package. The individual tutorials cover the design and implementation of your project.

• Group tutorials (for the Java agent package) will be at 14:00 on Mondays (starting 20th of February) in C60 (CS).
• Individual tutorials are scheduled for 15:00-16:00 on Thursdays in B37 (CS). Please email me to book a slot.

Assessment

Assessment is by project work and reports.

What is required

• An interim report containing the specification and outline design of your agents(s). The report should not exceed 2,000 words. Details of the content and format of the intermim report can be found in the Getting Started With Your Project lecture.
• A final report describing your agents(s) and the associated source code. This should consist of:
  1. a report describing your agents(s) - the report should not exceed 7,500 words excluding references;
  2. the code implementing your agents(s) (and the environment if you modified the standard environment or wrote your own).
  Some suggestions regarding the content and format of the final report can be found here. All reports should be submitted in pdf format -- submissions in other formats will not be marked.

How it will be marked

The aim of the module is to understand the relationship between an agent's task environment and its architecture. To do well, you must not only develop an agent (or agents) that work well, but demonstrate that you understand why it (or they) work well. Marking is therefore based on:

• the capabilities of the implemented agent system, including the quality of the specification, design and implementation;
• the degree to which the specification, design and implementation are clearly documented in the report, including any relevant background material used in the design and implementation of the agent(s); and
• clarity of presentation in general (including grammar, spelling and punctuation).

Very broadly, a basic implementation of the minimal requirements (and corresponding report) would gain a pass mark. Extra credit will be given for submissions that demonstrate a clear understanding of the relationship between the specified task environment and the architecture of the implemented system. Note that this does not necessarily involve implementing one of the extensions: it is possible to get a first class mark by doing an excellent implementation of the minimal requirements and an excellent report. The full assessment guidelines are available here.

You are reminded of the School's Policy on Plagiarism and the correct use of citations.

Submission dates

• Interim reports: 16:00 on Monday the 5th of March.
• Final reports and code: 16:00 on Tuesday the 27th of March.

All submissions should be made electonically using the coursework submission system. The ID for the final report and code is 484.

Recommended Reading

• Wooldridge (2002), An Introduction to MultiAgent Systems, Wiley.
• Russell & Norvig (2003), Artificial Intelligence: A Modern Approach, Prentice Hall (or the 1995 1st edition).
• Ferber (1999), Multi-Agent Systems: An Introduction to Distributed Artificial Intelligence, Addison-Wesley.
• Arkin (1998), Behavior-Based Robotics, MIT Press.
• Braitenberg (1984), Vehicles: Experiments in Synthetic Psychology, MIT Press.

Note that there is also a version of the reading list with links to the University Library catalogue.

Other Resources

• The AAAI Agents page has some overview papers and some useful links. The AAAI Springboard page also has lots of references to other AI topics, such as Machine Learning.
• The AI lab at the University of Michigan has a useful survey of cognitive and agent architectures.

Suggested Reading

Note that the following list of suggested reading is provisional, as it may necessary to change the order in which topics are presented.

• Introduction: Russell & Norvig (2003), chapter 1; Wooldridge (2002), chapter 2.
• Task Environments: Russell & Norvig (2003), chapter 2.
• Reactive Architectures I: Braitenberg (1984); see also the Braitenberg Vehicles simulator and Craig Reynolds Boids page.
• Reactive Architectures II: Arkin (1998), chapter 4.
• Deliberative Architectures I: Russell & Norvig (2003), chapter 11; Wooldridge (2002), chapter 4.
• Deliberative Architectures II: Russell & Norvig (2003), chapter 12 (or chapter 13 of the 1st edition); Wooldridge (2002), chapter 4. For the IRMA architecture, see Martha Pollack's papers, especially Pollack et al (1994) "Experimental Investigation of an Agent Commitment Strategy"; for Tileworld itself see the working paper by Mike Lees.
Project description: von Frisch (1966). -->
• Hybrid Architectures: Arkin (1998), chapter 6; Wooldridge (2000), chapter 5, section 5.3. Xavier the robot
• Multi-Agent Systems I: Ferber (1999), chapters 1 and 4.
• Multi-Agent Systems II: Wooldridge (2003), chapter 9.

Lecture Slides

• Lecture 1: Introduction Slides The Wired video of the winner of the DARPA Urban Challenge 2007 can be found here.
• Lecture 2: Task Environments Slides
• Lecture 3: Reactive Architectures I Slides See also the Braitenberg Vehicles simulator and Craig Reynolds Boids page.
• Lecture 4: Reactive Architectures II Slides
• Lecture 5: Project Description Slides
• Lecture 6: Getting Started With Your Project Slides (with corrected submission date for final reports)
• Lecture 7: Deliberative Architectures I Slides. See also the travelling salesman applet and the Shakey video (note that the video is about 100MB).
• Lecture 8: Deliberative Architectures II Slides
• Lecture 9: Hybrid architectures Slides (including Module evaluation feedback). The Xavier videos are available here: environment, planning, and elevator.
• Lecture 10: Multi-agent systems Slides

---

Copyright © 2012 Brian Logan

This file is maintained by Brian Logan

Last modified: 31-Jan-2012, 11:03