artificial intelligence chapter 1: Game AI

Download Report

Transcript artificial intelligence chapter 1: Game AI

Alexander Repenning
artificial intelligence
chapter 1: Game AI
Objectives
•
learn about difference between AI and Game AI
•
learn about a new AI approach called Collaborate
Diffusion
game AI
•
single Agent
•
ALife: agent acts intelligent: develops goals based
on needs, pursues goals.
•
path finding (e.g., A*):
•
•
•
artificial opponents finds ways trough maze to
get you
•
Sims: find refrigerator in house and food inside
learning: artificial opponents learn about your
behavior making game play progressively harder
multi Agents
•
flocking, emergence
•
collaboration
challenges
•
•
Computational:
•
AI needs to “run” at 60 frames per second
•
symbolic AI is (mostly) non-incremental
Psychological:
•
AI needs to “look” right
•
often very simple, e.g., random, e.g. Mt. Vetro’s
eyes
more pointers:
•
good site: http://www.gameai.com/
•
new book: AI for Game Developers, David M. Bourg
how to track Pacman?
ideas
•
Diffusion Search: combine the notion of diffusion (a
formal conceptualization on how things spread) with
Search, e.g., hill climbing
•
Collaborate Diffusion: use Diffusion Search in a multi
agent setting to express collaboration and competition
diffusion
(physics) the process of diffusing; the intermingling of molecules in gases and liquids as a result of
random thermal agitation
www.cogsci.princeton.edu/cgi-bin/webwn
the spread of social institutions (and myths and skills) from one society to another
www.cogsci.princeton.edu/cgi-bin/webwn
dissemination: the property of being diffused or dispersed
www.cogsci.princeton.edu/cgi-bin/webwn
dispersion: the act of dispersing or diffusing something; "the dispersion of the troops"; "the diffusion
of knowledge"
www.cogsci.princeton.edu/cgi-bin/webwn
The movement of chemical species (ions or molecules ) under the influence of concentration
difference. The species will move from the high concentration area to the low concentration area till
the concentration is uniform in the whole phase. Diffusion in solutions is the most important
phenomenon in electrochemistry, but diffusion will occur also in gases and solids.
electrochem.cwru.edu/ed/dict.htm
the movement of particles from an area of higher concentration to an area of lower concentration
coris.noaa.gov/glossary/glossary_a_k.html
Collaborative
Diffusion
•
well suited for complex, multi-agent simulation game:
path finding, ALife, flocking, emergence and
collaboration
•
new: developed at CU, started on Connection Machine
•
computationally expensive but at the same time
incremental: works well on current computers and as
part of game engines
•
traditional game AI (e.g., A* for pathfinding)
approaches are not incremental
characteristics
•
Spatial Extend: works for agents with spatial
relationships (2D, 3D, connection machine: 12D)
•
Simple to Program: algorithms are computationally
expensive but relatively simple to built and tweak.
•
Ecological
•
traditional AI: AI in agent, e.g., robot
•
distributed AI: AI in agents ⇒ flocking...
•
ecological AI: AI everywhere: agents & environment
•
Parallel: no chess-like turn taking
•
Incremental: AI state is part of environment and
continuously updated
•
Robust: likely to work with situations not anticipated,
e.g., soccer with n goals, m balls for n, m ≠ 2
diffusion
equation
u2
u1
u0
u3
u4
n
u0 (t  1)  u0  D (ui  u0 )
i1
•
u0 = D (u1 + u2 +u3 +u4 - 4u0) + u0
•
D: Diffusion coefficient [0..0.5]
•
simple: D = 0.25 => u0 = 0.25 *(u1 + u2 + u3 + u4)

4) Collaborative
Problem Solving
•
multiple collaborative agents
•
collaborating: soccer, players from the same team
•
competing: soccer, players from the other team
•
changing goals: first track ball, then kick ball into goal
•
simple version: Collaboration trough Goal Obfuscation
World Cup
sample projects
•
MySims: a version of the Sims
•
The Madness of Crowds: how people behave in panic