DeepAI AI Chat
Log In Sign Up

Learning to Coordinate with Humans using Action Features

by   Mingwei Ma, et al.

An unaddressed challenge in human-AI coordination is to enable AI agents to exploit the semantic relationships between the features of actions and the features of observations. Humans take advantage of these relationships in highly intuitive ways. For instance, in the absence of a shared language, we might point to the object we desire or hold up our fingers to indicate how many objects we want. To address this challenge, we investigate the effect of network architecture on the propensity of learning algorithms to exploit these semantic relationships. Across a procedurally generated coordination task, we find that attention-based architectures that jointly process a featurized representation of observations and actions have a better inductive bias for zero-shot coordination. Through fine-grained evaluation and scenario analysis, we show that the resulting policies are human-interpretable. Moreover, such agents coordinate with people without training on any human data.


Deep learning control of artificial avatars in group coordination tasks

In many joint-action scenarios, humans and robots have to coordinate the...

PECAN: Leveraging Policy Ensemble for Context-Aware Zero-Shot Human-AI Coordination

Zero-shot human-AI coordination holds the promise of collaborating with ...

On the Utility of Learning about Humans for Human-AI Coordination

While we would like agents that can coordinate with humans, current algo...

Adaptive Coordination in Social Embodied Rearrangement

We present the task of "Social Rearrangement", consisting of cooperative...

Equivariant Networks for Zero-Shot Coordination

Successful coordination in Dec-POMDPs requires agents to adopt robust st...

A New Formalism, Method and Open Issues for Zero-Shot Coordination

In many coordination problems, independently reasoning humans are able t...