CREW: Platform for Human-AI Teaming
Introduction
CREW is a platform designed to facilitate Human-AI teaming research, engage collaborations from multiple scientific disciplines, with a strong emphasis on human involvement. It includes pre-built tasks for cognitive studies and Human-AI teaming with expandable potentials from our modular design. Following conventional cognitive neuroscience research, CREW also supports multimodal human physiological signal recording for behavior analysis. Moreover, CREW benchmarks real-time human-guided reinforcement learning agents using state-of-the-art algorithms and well-tuned baselines.
To get started, check out Getting Started for basic examples and Tutorials for advanced usage.
Video (click to play on YouTube)
Overview of Platform
CREW consists of two main subcomponents: Dojo and Algorithms.
These subcomponents work together to create an efficient platform for developers and researchers alike.
Dojo serves as a Unity package designed specifically to facilitate the development of multiplayer games involving human and AI players. We provide a set of pre-built environments as well as a template for building custom tasks with real-time interaction enabled.
Algorithms, on the other hand, is a Python package aimed at researchers who wish to create AI agents capable of operating and collaborating with humans within the environments established by Dojo. Offering an intuitive interface, Algorithms ensures maximum flexibility and customizability for the researchers.
By working in unison, these two subcomponents create a robust and user-friendly platform for the development of interactive experiences.
Authors
CREW is a fully open-sourced project developed by General Robotics Lab at Duke University.
Funding Acknowledgement
This work is supported in part by ARL STRONG program under awards W911NF2320182 and W911NF2220113.