RTSCup: testbed for multiagent systems evaluation

AUTOR(ES)

Vicente Vieira Filho

DATA DE PUBLICAÇÃO

2008

RESUMO

The evaluation of any computational system is an important phase of its development process and this is not different to systems that use Artificial Intelligence (AI). For such systems, in particular, there is a trend in employing simulation environments as testbeds, which enable the evaluation of such systems in different test scenarios. The area of interest of this work is Multiagents Systems (MAS). This area of research is booming because MAS are being used to solve problems which are difficult or impossible for an individual agent or monolithic system to solve. Besides, there are many interesting problems that arise during the interactions among agents which often involve distributed problem solving on the fly (real-time). There are many testbeds used to evaluate Multiagent Systems such as Trading Agent Competition, RoboCup Rescue and ORTS. However most of these testbeds do not have the appropriate features to help researchers to define, implement and validate their hypothesis. Most features not addressed by simulators are related to usability and software engineering aspects. The aim of this work is to present a simulator to Multiagent Systems, called RTSCup, which can be used as a testbed for evaluating several AI techniques used to implement teams as Multiagent Systems (MAS). RTSCup was already employed in a practical experiment during a competition among AI students of the Federal University of Pernambuco.

ASSUNTO(S)

estratégia em tempo real benchmark testbed jogos eletrônicos testbed sistemas multiagentes benchmark ciencia da computacao multiagent systems real-time strategy eletronic entertainment

Documentos Relacionados

• Governing multi-agent systems
• Systems of exchange values as tools for multi-agent organizations
• A testbed for network performance evaluation and its application to connection admission control algorithms
• An integrated role-based approach for modeling, designing and implementing multi-agent systems
• Engineering multi-agent systems with aspects and patterns