multimedia @ VU

---

[] readme titles system demo dlp vr slides resources

---

3D Agent-based Virtual Communities

agents

3D Agent-based

Virtual Communities

Zhisheng Huang

Anton Eliens

Cees Visser

Vrije Universiteit, Amsterdam

structure

• background -- the WASP/RIF project
• agents in virtual worlds -- agent communication
• programming platform -- DLP+VRML
• case study -- multi-user soccer
• conclusions and future research

WASP

Web Agent Support Program

RIF

Retrieval of Information

in virtual worlds

using Feature detectors

RIF + WASP

• distributed logic programming -- uniform platform
• agent technology -- subsumes multi-user server

agents in virtual environments

• virtual environments with embedded autonomous agents
• virtual environments supported by ACL communication

ACL = Agent Communication Language

shared objects

• pilot -- instance that will be replicated
• drone -- instance that replicates pilot

• pilot agents -- control state of a shared object
• drone agents -- replicate the state of a shared object

agents

• object agents -- controls a single shared object (like the soccer ball)
  pilot at server, drone at client
• user agent -- controls users' avatar
  pilot at user side, drone at server or clients
• autonomous agents -- like football player, with own avatar
  pilot at server, drone at clients

programming platform

• VRML EAI support
• distributed communication capabilities (TCP/IP)
• multiple threads of control -- for multiple shared objects
• declarative language -- for agent support

DLP

Distributed Logic Programming

control points

• get/set -- position, rotation, viewpoint

  findHowToReact(Agent,Ball,Goal,shooting) :-
     get(Agent,position,sfvec3f(X,Y,Z)),
     get(Ball,position,sfvec3f(Xb,Yb,Zb)),
     get(Goal,position,sfvec3f(Xg,Yg,Zg)),
     distance(sfvec3f(X,Y,Z),sfvec3f(Xb,Yb,Zb),DistB),
     distance(sfvec3f(X,Y,Z),sfvec3f(Xg,Yg,Zg),DistG),
     DistB =< kickableDistance,
     DistG =< kickableGoalDistance.
  

multi-user soccer game

• multiple (human) users -- may join during the game
• multiple agents -- to participate in the game (e.g. goalkeeper)
• reactivity -- players (users and agents) have to react quickly
• cooperation/competition -- requires intelligent communication
• dynamic behavior -- sufficiently complex 3D scenes, including the dynamic behavior of the ball

football

END OF GAME powered by DLP

cognitive loop

• sense -- think -- act

roles

• goal keeper
• defender
• mid-fielder
• forward

condition

• field coverage

distributed game protocol

[ Act, Type, Parameters ]

• register game: [ register, game_name, from(Host) ]
• tell new player: [ tell, new_player, user(Host,Name) ]
• tell kick ball: [ tell, kick_ball, [ user(Host,Name), force(X,Y,Z) ] ]
• ask game score: [ ask, game_score, user(Host,Name) ]

register accept, register wait, tell position, tell rotation, tell game score, reply game score, unregister game, reply unregister game, player gone ...

performance improvement

• compound messages: run and trace

conclusions

virtual environments with intelligent agents offer a general solution to shared objects and autonomous creatures in VRML worlds.

DLP+VRML

pro

• high level platform -- flexible and powerful
• clean separation of (modelling and programming) concerns

con

• added complexity
• (some) performance penalty

future work

• animation of faces and avatars
• ....

www.cs.vu.nl/~eliens/research

---

[] readme titles system demo dlp vr slides resources

---