Thesis (Selection of subject)

Your browser does not support JavaScript, or its support is disabled. Some features may not be available.

StarCraft and Emergent in Pogamut 3 environment

Thesis title in Czech:	Pogamut a StarCraft v prostředí Emergent
Thesis title in English:	StarCraft and Emergent in Pogamut 3 environment
Key words:	virtual agent, neural network, action selection mechanism, game
English key words:	virtual agent, neural network, action selection mechanism, game
Academic year of topic announcement:	2011/2012
Thesis type:	diploma thesis
Thesis language:	angličtina
Department:	Department of Software and Computer Science Education (32-KSVI)
Supervisor:	doc. Mgr. Cyril Brom, Ph.D.
Author:	hidden - assigned and confirmed by the Study Dept.
Date of registration:	03.11.2011
Date of assignment:	04.11.2011
Confirmed by Study dept. on:	29.11.2011
Date and time of defence:	09.09.2014 10:00
Date of electronic submission:	31.07.2014
Date of submission of printed version:	31.07.2014
Date of proceeded defence:	09.09.2014
Opponents:	Mgr. et Mgr. Filip Děchtěrenko, Ph.D.

Guidelines

Pogamut 3 je Netbeans plugin určený k rychlému programování chování racionálních agentů. Doposud byl používán primárně k programování agentů pro 3D prostředí hry Unreal Tournament 2004 (dále jen UT2004) a testování symbolických rozhodovacích technik agentů. Cílem práce je ověřit flexibilitu nástroje Pogamut 3, konkrétně 1) zdali je možné ho připojit k výrazně odlišnému hernímu prostředí a 2) zdali je možné ho použít k testování agentů s konekcionistickými rozhodovacími mechanismy.

Možnost připojit Pogamut 3 k typově rozdílným prostředím bude ověřeno na propojení s realtimovou strategickou hrou Starcraft : Broodwars (SCBW). V případě úspěchu tohoto propojení student umožní programovat inteligentní agenty pro SCBW v prostředí Pogamut 3. V rámci možností budou také přizpůsobeny stávajících moduly Pogamutu 3 směrem k jejich použitelnosti pro tvorbu agentů určených pro SCBW. Toto rozšíření bude uzavřeno proof-of-concept implementací umělé inteligence, provedené v Pogamutu, která bude na jednoduché úrovni hrát SCBW za některou ze tří ve hře dostupných ras.

Možnost využití konekcionistických agentů bude otestována na propojení Pogamutu 3 s nástrojem pro simulování komplexních neuronových sítí a kognitivních procesů Emergent (dříve PDP++) a prostředí hry UT2004. V ideálním případě bude tato část práce zakončena implementací jednoduchého proof-of-concept UT2004 agenta v Pogamutu 3, na jehož zrakovém vnímání se bude podílet neuronová síť simulovaná v Emergentu.

References

Jilk, D.J., Lebiere, C., O'Reilly, R.C. & Anderson, J.R. (2008). SAL: An explicitly pluralistic cognitive architecture. Journal of Experimental and Theoretical Artificial Intelligence, 20, 197-218.

O'Reilly, R.C., Munakata Y. (2000) Computational Explorations in Cognitive Neuroscience: Understanding the Mind by Simulating the Brain, MIT Press.

Gemrot, J., Kadlec, R., Bida, M., Burkert, O., Pibil, R., Havlicek, J., Zemcak, L., Simlovic, J., Vansa, R., Stolba, M., Plch, T., Brom C. (2009) Pogamut 3 Can Assist Developers in Building AI (Not Only) for Their Videogame Agents. In: Agents for Games and Simulations, LNCS 5920, Springer, pp. 1-15.

J. Gemrot, C. Brom, R. Kadlec, M. Bida, O. Burkert, M. Zemcak, R. Pibil. (2010) Pogamut 3 - Virtual Characters Made Simple. In: Advances in Cognitive Science, Gray, J. eds, The Institution Of Engineering And Technology, pp 211-243.

Preliminary scope of work

Pogamut 3 is a plugin for the Netbeans IDE designed for rapid programming of rational agents behavior. Up to the present time it was used primarily for programming agents for the 3D environment of the computer game Unreal Tournament 2004 (hereinafter referred to as UT2004) and testing the symbolic decision techniques of agents.

The goal of this thesis is to test the flexibility of Pogamut 3, namely 1) whether it is possible to connect it to a considerably different game environment and 2) whether it is possible to utilize it for testing of agents with connectionist decision-making mechanisms.

The option of connecting Pogamut 3 to a substantially different environment will be tested by connecting it to the real-time strategic game StarCraft: Broodwars (hereinafter referred to as SCBW). In the event of success of this interconnection, student will allow for intelligent SCBW agents to be coded in the Pogamut 3 environment. Also existing Pogamut 3 modules will be adapted within the bounds of possibility towards their usability for the creation of SCBW agents. This Pogamut 3 expansion will be wrapped up by a proof-of-concept implementation of the artificial intelligence in Pogamut 3, able to play SCBW on a simple level for one of the three game races available in the SCBW.

The possibility of using Pogamut 3 for coding the connectionist agents will be tested on the interconnection of Pogamut 3 and the tool designed for the simulation of complex neural networks and cognitive processes Emergent (formerly PDP++) in the UT2004 game environment. Ideally this part of work will be ended by a proof-of-concept implementation of such simple UT2004 agent in Pogamut 3, to whose visual perception a neural network, simulated in Emergent, will contribute.

Preliminary scope of work in English

Pogamut 3 is a plugin for the Netbeans IDE designed for rapid programming of rational agents behavior. Up to the present time it was used primarily for programming agents for the 3D environment of the computer game Unreal Tournament 2004 (UT2004) and testing the symbolic decision techniques of agents.

The goal of this thesis is to test the flexibility of Pogamut 3, namely 1) whether it is possible to connect it to a considerably different game environment and 2) whether it is possible to utilize it for testing of agents with connectionist decision-making mechanisms.

The option of connecting Pogamut 3 to a substantially different environment will be tested by connecting it to the real-time strategic game StarCraft: Broodwars (SCBW). In the event of success of this interconnection, the student will allow for intelligent SCBW agents to be coded in the Pogamut 3 environment. Also existing Pogamut 3 modules will be adapted within the bounds of possibility towards their usability for the creation of SCBW agents. This Pogamut 3 expansion will be wrapped up by a proof-of-concept implementation of the artificial intelligence in Pogamut 3, able to play SCBW on a simple level for one of the three game races available in the SCBW.

The possibility of using Pogamut 3 for coding the connectionist agents will be tested on the interconnection of Pogamut 3 and the tool designed for the simulation of complex neural networks and cognitive processes Emergent (formerly PDP++) in the UT2004 game environment. Ideally this part of work will be ended by a proof-of-concept implementation of such simple UT2004 agent in Pogamut 3, to whose visual perception a neural network, simulated in Emergent, will contribute.