Learning from Observations Using a Single Video Demonstration and Human Feedback

TL;DR

This paper introduces a method for training agents from a single video demonstration by using human feedback to align visual and standard state representations, enabling effective learning and transfer to new tasks.

Contribution

The novel approach combines video demonstrations with human feedback to map visual and standard representations, facilitating learning from a single demonstration and transfer to new tasks.

Findings

Successfully trained a hopper to perform a backflip from a single video.

Demonstrated transfer learning to new tasks with minimal human feedback.

Effective alignment of visual and standard representations using neural networks.

Abstract

In this paper, we present a method for learning from video demonstrations by using human feedback to construct a mapping between the standard representation of the agent and the visual representation of the demonstration. In this way, we leverage the advantages of both these representations, i.e., we learn the policy using standard state representations, but are able to specify the expected behavior using video demonstration. We train an autonomous agent using a single video demonstration and use human feedback (using numerical similarity rating) to map the standard representation to the visual representation with a neural network. We show the effectiveness of our method by teaching a hopper agent in the MuJoCo to perform a backflip using a single video demonstration generated in MuJoCo as well as from a real-world YouTube video of a person performing a backflip. Additionally, we show that our method can transfer to new tasks, such as hopping, with very little human feedback.