Transcript 4 - Computational Intelligence Lab

Poster Presentation Code: THU-P1-01
School of Computer Engineering
The 3rd Conference on
Industrial Electronics and Applications
Behavior Decision Algorithm Using SGNN
for Game Characters
Background
Experiment
 Game artificial intelligence refers to techniques
that make computer and video games to produce
the illusion of intelligence in the behavior of nonplayer characters (NPCs).
 The behavior structure of the NPCs is often
crucial in the game, since players are already
tired about the simple arranged, well-regulated
and even stiff actions of the non player creatures.
 Game agent is in the need of improvement to
meet the increasing demand of game players.
 Method is applied in computer games to
extract the non-player characters’ behavior
logic rules based on human knowledge and
experience,
make
the
NPCs
active
reasonable and more like real human beings,
and contribute to enhance computer games
interest and intelligence ability.
 For example, in military games, a NPC
soldier’s behavior which maybe include shoot,
guard, move and chase are affected by
environment
parameters
and
internal
attributes.
Methodology
 Our research focus on improving fuzzy logic,
neural network, and other related techniques and
apply these artificial intelligent computing
methods to model the non player characters.
 A chaotic behavior decision algorithm is
proposed based on self-generating neural
network and fuzzy logic.
 Offline Behavior Rule Extraction based on
SGNN
 Online Fuzzy Behavior Decision Algorithm
Fig3. Classified behaviors using SGNN
Human Expert
Players Experience
Online Data of NPC

x
Fig1.
Self-generated neural
network
Contributors:
SGNN Generation
Online Behavior
Decision
Offline Rule
Extraction
Rule Base
N
P
C
Fig2.
Fuzzy function used
Fig 4. Flow chart of chaotic behavior decision.
Note how the NPCs behavior is decided during
progress
Feng Shu; Narendra S. Chaudhari
www.ntu.edu.sg