class: center, middle
Gerard Escudero, 2020
.footnote[Source: inverse ]
class: left, middle, inverse
-
.cyan[Artificial Intelligence]
-
Game AI
-
Content & Evaluation
-
References & Tools
-
an agent does tasks normally attached to humans (or animals)
-
an agent has AI when it exhibits "smart" features:
- learning, reasoning, behavior, reaction
-
play chess
-
reading license plates
-
weather forecasting
.cols5050[ .col1[
-
Early Days: phylosofy
Turing test (1950) -
1950: symbolic era
Knowledge set + resoning -
1960: cybernetics
biological (neural networks) -
1966: crisis!
ALPAC report against MT -
1970: cybernetics (MLP)
symbolic (Search Strategies, Uncertainty) -
1980: real applications
expert systems, fuzzy logic ] .col2[ -
1990-2000: machine learning (statistical)
-
2010-: deep learning ]]
class: left, middle, inverse
-
.brown[Artificial Intelligence]
-
.cyan[Game AI]
-
Content & Evaluation
-
References & Tools
-
1980: Pac-Man (state machine)
-
1994: Warcraft (pathfinding)
-
1997: Goldeneye 007 (sense simulation)
-
1997: Creatures (AI centered game)
-
1998: Warhammer: Dark Omen (formation motion)
-
1998: Half-Life (AI collaborators)
-
2000: The Sims (AI centered game)
-
2001: Halo (Decision Trees)
-
2001: Black & White (AI centered game)
-
2005: F.E.A.R. (GOAP)
-
2015: AlphaGo (Deep Learning)
Building a game experience should be our main objective.
.blue[The Complexity Fallacy]:
-
Select always the simplest way to give the illusion of complexity
-
.blue[Occam's Razor]: the simplest explanation is most likely the right one
.blue[Perception window]:
-
Interaction with agents lasts a very short time.
-
Try to avoid commonsense behavioral changes
(see 2on video at references).
.blue[Memory & speed restrictions]
class: left, middle, inverse
-
.brown[Artificial Intelligence]
-
.brown[Game AI]
-
.cyan[Content & Evaluation]
-
References & Tools
.blue[Main concepts]:
-
Movement (steering behaviours)
-
Planning for movement (pathfinding)
-
Decision Taking (behavior trees)
-
Machine Learning (neural networks and deep learning)
-
AI Game Design
.blue[Goals]:
-
Game AI: common tools
-
Unity & C#
-
.blue[Steering Behaviours]
- Wander & Hide
- Flocking
-
.blue[Navigation Mesh & Pathfinding]
-
.blue[World Interface] (senses)
-
.blue[Decision Making]
- Behaviour Trees
- Smart Objects & Planners
-
.blue[Group tactical movement and decision taking]
- Tactical group movement & Blackboards for information sharing
- Behavior Trees for group coordination
-
.blue[Machine Learning]
- Deep Learning
- Reinforced Learning
.cols5050[ .col1[
- Counts as 20%
- November 1st at 23:59
- Movement: steerings & pathfinding
- Counts as 20%
- December 8th at 23:59
- Behaviour: Behaviour Trees
- Counts as 25%
- January 10th at 23:59
- Machine Learning ] .col2[
- Work is in teams of 2
- Counts as 10%
- Counts as 25%
- January
- Counts as 25%
- Max grade is 5
- January 30th - February 5th ]]
class: left, middle, inverse
-
.brown[Artificial Intelligence]
-
.brown[Game AI]
-
.brown[Content & Evaluation]
-
.cyan[References & Tools]
- Ian Millington. AI for Games (3rd ed). CRC Press, 2019.
-
Unity (personal edition)
-
Github (private repositories)
.blue[See next videos]:
- Artificial Intelligence: Skyrim - The Elder Scrolls V. Ether Dynamics,
-
Stupidest AI Bug Contest 2011. Masi84, 2011.
-
Top 10 Video Games with the Best AI. WatchMojo.com, 2015.
-
MarI/O - Machine Learning for Video Games. SethBling, 2015.
-
The wonderful and terrifying implications of computers that can learn. Jeremy Howard, 2014.