Adapt2Reward: Adapting Video-Language Models to Generalizable Robotic Rewards via Failure Prompts

TL;DR

This paper introduces Adapt2Reward, a method that leverages failure prompts and clustering of failure videos to create a generalizable language-conditioned reward function for robots, enabling better adaptation to new environments and instructions.

Contribution

It presents a novel approach to transfer vision-language models into reward functions using minimal task data and failure clustering, improving robotic generalization.

Findings

Outperforms existing reward models in new environments

Effective in distinguishing success and failure modes

Enhances robot planning and reinforcement learning

Abstract

For a general-purpose robot to operate in reality, executing a broad range of instructions across various environments is imperative. Central to the reinforcement learning and planning for such robotic agents is a generalizable reward function. Recent advances in vision-language models, such as CLIP, have shown remarkable performance in the domain of deep learning, paving the way for open-domain visual recognition. However, collecting data on robots executing various language instructions across multiple environments remains a challenge. This paper aims to transfer video-language models with robust generalization into a generalizable language-conditioned reward function, only utilizing robot video data from a minimal amount of tasks in a singular environment. Unlike common robotic datasets used for training reward functions, human video-language datasets rarely contain trivial failure videos. To enhance the model's ability to distinguish between successful and failed robot executions, we cluster failure video features to enable the model to identify patterns within. For each cluster, we integrate a newly trained failure prompt into the text encoder to represent the corresponding failure mode. Our language-conditioned reward function shows outstanding generalization to new environments and new instructions for robot planning and reinforcement learning.