Syllabus
Lecture topics:
1. Taxonomy of artificial beings and their applications: learning simulations, video games, serious games, virtual storytelling, interactive drama, computational ethology.
2. Symbolic approaches to action selection: reactive planning, deliberative methods; if-then rules, finite-state machnies, behavioral trees, subsumption, Belief-Desire-Intention architecture, multi-layered architetures.
3. Connectionist approaches to action selection: free-flow hierarchies (Tyrrell), neural networks (Creatures, Black&White), approaches to agent learning.
4. Introduction to ethology: Psychohydraulic model of Konrad Lorenz, models of population dynamics.
5. Path-planning: steering rules, A*, HPA*.
6. Environment representation: affordances, smart objects, nav-mesh, way-points, sensory versimilitude.
7. Memory: psychological classification, short-term memory & episodic memory for the agents.
8. Unified theories of cognition: SOAR, ACT-R Practical lessons are carried out in virtual environments of two different video games: Unreal Tournament 2004 (UT2004) and NOTA. Practical lesson topics in UT2004:
1. Revisiting Java (sytax, collections, lists, sets, maps, iterators, lazy initialization, observer pattern and its problems, weak references), Maven basics
2. Introduction to the Pogamut platform and virtual environment of UT2004
3. Events and objects in Pogamut, listeners, annotations.
4. BOD methodology for the behavior design of virtual agents; if-then rules, finite state machines, behavior breakdown
5. Steerings and low-level movement of bots
6. Navigation in UT2004
7. Visibility in UT2004, non-trivial use of A* Practical lesson topics in NOTA:
1. Introduction to Lua scripting language
2. Behavior trees
3. Behavior patterns in behavior trees
4. Controlling groups of bots using behavior trees
Annotation
In this course, we will study human-like artificial agents, that is autonomous intelligent agents situated in a virtual environment similar to real world that act like humans. The course gives an overview of types of such agents and their architectures with the emphasis on the problem of action selection.
The course also focuses on solving practical issues related to real-time and partially observable environments.