Trust Your Agents

Overview

We are working on the “Trust your agents” project, which aims to develop a novel approach to the collaboration between humans and machines. The project is funded by the Air Force Office of Scientific Research (AFOSR) under the Trust and Influence program (RTA2-8). The goal is to show that human-AI collaboration can be more effective than AI-only or human-only systems when it comes to completing a collaborative task.

Our approach focuses on developing novel models and algorithms for model-based and hierarchical multi-agent deep reinforcement learning. The hierarchical decomposition of a high-complex scenario allows the agents to reason on multiple levels of abstraction and to share only coarse-grained information with human collaborators while still being able to share more fine-grained or low-level information with machine collaborators. By using appropriate and interpretable communication systems, such as images, text, or audio, the agents can transmit their internal representations to human teammates, who can thus acquire greater confidence in the abilities of their machine partner.

We aim to develop trust between humans and artificial teammates by using the prediction capabilities of the model-based approach combined with explainable communication. The learned models can also help in explaining, monitoring, and possibly driving the factors that most influence human behaviors.