220 likes | 233 Views
Learn about ORTS, an open source real-time strategy game engine used for studying real-time AI problems. Participate in the ORTS RTS Game AI Competition and learn how to run and study ORTS. Explore topics like multi-agent systems, planning, and tactics. Find additional resources for further study.
E N D
ORTS A Free Software RTS Game Engine
Conversa de Hoje • O que é ORTS? • Que competição é essa? • Game 1 • Game 2 • Game 3 • Como rodar? • Como estudar?
O que é ORTS • ORTS = Open Real-Time Strategy • É um ambiente de programação para o estudo de problemas de real-time IA tais como: • Pathfinding • Imperfect information • Scheduling • Planejamento
Que competição é essa? • ORTS RTS Game AI Competition • A competição envolve o uso da Engine ORTS e é dividida em categorias: • Game 1 :: “Gathering Resources” • Game 2 :: “Tank Combat” • Game 3 :: “A Real RTS Game”
Game 1 • Objetivo • Gather as much resources as possible within 10 minutes • Desafios • Cooperative pathfinding • static/mobile obstacle avoidance
Game 1 • Setup • single player • perfect information • random 32x32 terrain (ground level tiles connected, small irregular static obstacles) • one control center, 20 workers nearby • several resource patches • limit # of workers on single resource patch to 4 • some small mobile obstacles ("sheep") moving randomly
Game 1 • Ações - (Worker) • move(x,y[,s]) • stop() • pickaxe.mine(obj) • return_resources(obj)
Game 2 • Objetivo • Destroy as many opponent buildings as possible within 15 minutes • Desafios • small-scale combat • unit group management • adversarial/cooperative pathfinding
Game 2 • Setup • two players • perfect information (apart from simultaneous actions) • random 64x64 terrain (ground level connected, small irregular static obstacles) • for each player: 5 randomly (but symmetrically) located control centers with 10 tanks each nearby • some small mobile obstacles ("sheep") moving randomly
Game 2 • Ações – (Tank) • move(x,y[,s]) • weapon.attack(obj) • stop()
Game 3 • Objetivo • destroy all opponent buildings within 20 minutes • Desafios • pathfinding • combat • grouping forces • scouting • resource allocation
Game 3 • Setup • two players • imperfect information (simultaneous actions + fog of war) • random 64x64 terrain (ground level connected, small irregular static obstacles) • some neutral mobile obstacles ("sheep"), random motion • one random starting location each with one control center, a resource cluster and 6 workers nearby • ~4 more resource locations randomly scattered on terrain • tech-tree: • workers build control centers, barracks, and factories • control centers train workers • barracks train marines • factories produce tanks • building a barracks requires control center • building a factory requires control center and barracks
Game 3 • Ações • Worker • move(x,y[,s]) • weapon.attack(obj) • stop() • pickaxe.mine(obj) • return_resources(obj) • build_controlCenter(x,y) • build_barracks(x,y) • build_factory(x,y)
Game 3 • Ações • Tank • move(x,y[,s]) • weapon.attack(obj) • stop() • Marine • move(x,y[,s]) • weapon.attack(obj) • stop()
Game 3 • Ações • Control Center • train_worker () • stop() • Barracks • train_marine() • stop() • Factory • build_tank() • stop()
Como rodar? • No site da competição e do próprio ORTS possui informações e how-to’s • ORTS • http://www.cs.ualberta.ca/~mburo/orts/index.html • ORTS Competition • http://www.cs.ualberta.ca/~mburo/orts/AIIDE06/
Como estudar? • Tópicos importantes • Multi-agent • Foraging • Planning • Formação • Táticas & Estratégias • ...
Como estudar? • AI Game Wisdom 1 e 2 • Game Programming Gems 1, 2 e 3 • Steels, L. (1989). Cooperation between Distributed Agents through Self-Organisation. Decentralized A.I. E. Werner and Y. Demazeau. Amsterdam, Elsevier North-Holland.
ORTS Dúvidas?