15-396: Modeling of Cognitive Agents

Spring 03

Instructor: Scott Douglass
Office Hours: BH-4461, Fri. 1:00 to 3:00PM (or by appointment)

TA: Roman Stanchak (`roman@andrew.cmu.edu`)
Office Hours: TBA

Lecture

Where: NSH 1305
When: Tue & Thu 9:00 to 10:20AM

Lab

Where: WH 5336
When: Wed 9:30 to 10:20 AM

In a Nutshell:

This course will introduce mainstream computer science students to cognitive modeling. The course will present an agent behavior specification formalism based on first-order predicate calculus, behavioral models, and patterns of action. Students will use this formalism to explore how models of human behavior and cognition can be used to build psychologically plausible and engaging agents. Agent behaviors such as steering, navigation, simple communication, and qualitative spatial reasoning will be developed and assessed within a virtual environment.

Course Materials:

Course Information:

Texts & Required Purchases:

Multi-Agent Systems: An Introduction to Distributed Artificial Intelligence, by Jacques Ferber
UML Distilled: 2nd Edition, by Martin Fowler & Kendall Scott
Unreal Tournament (~$10-20 from most software shops)

Getting files:

Required files can be directly downloaded via:(HTTP)

Example Agents:

Agents-1-4: ( UML), ( 1, 2, 3, 4) Lisp -- collection of agents 1 throught 4

Rational Rose was used to specify agents in stereotype-extended UML. Petal format files from Rose were automatically converted into common lisp sources thatare directly supported by an agent deployment/simulation framework. The simulation frameworkuses FOPC internally and FOPC functions (n+1 arity predicates) can be used to "callout" tocommon lisp when conventional computations are required (see agents 4 and 5 for callout demonstrations).

Note: Older browsers with antiquated versions of Java will NOT be able to render the UML models.