Home
Welcome! Here you will find everything you need to know about running and experimenting with the Metaverse framework.
The Metaverse repository is dedicated to the development of a framework to support the empirical evaluation of cognitive agents in a variety of virtual environments. The framework is divided into two interrelated sub-systems; the MetaMind, and the Multiverse; each sub-system controls various aspects required to orchestrate various modular components.
MetaMind is an abstract interface for the creation, customization, and evaluation of intelligent cognitive agents based on architectures conforming to the Common Model of Cognition [ref]. Supported architectures currently include:
- ACT-R
- SOAR
- Sigma (coming soon)
Multiverse is an abstract interface for a multitude of virtual environments that support the training and testing of intelligent agents.
Supported environments:
- OpenAI Gym (Single agent, perfect information)
- Starcraft II (Single agent, imperfect information)
- MITRE Caldera (Red/Blue team)
- HELK/Mordor (coming soon)
These two subsystems are tied together by a common graphical interface to run experiments with various combinations of architectures and environments, and to measure the interface points within each agent, and the overall effects on external performance. The interface is based on PyQt5.