This&That: Language-Gesture Controlled Video Generation for Robot Planning

TL;DR

This paper introduces This&That, a framework that uses language and gesture-conditioned video generation to improve robot planning and execution of complex tasks, outperforming previous methods.

Contribution

It presents a novel approach combining video generative models with language-gesture conditioning for clearer task communication and improved robot planning.

Findings

Outperforms prior state-of-the-art behavior cloning methods

Enables unambiguous task communication with simple instructions

Successfully translates visual plans into robot actions

Abstract

Clear, interpretable instructions are invaluable when attempting any complex task. Good instructions help to clarify the task and even anticipate the steps needed to solve it. In this work, we propose a robot learning framework for communicating, planning, and executing a wide range of tasks, dubbed This&That. This&That solves general tasks by leveraging video generative models, which, through training on internet-scale data, contain rich physical and semantic context. In this work, we tackle three fundamental challenges in video-based planning: 1) unambiguous task communication with simple human instructions, 2) controllable video generation that respects user intent, and 3) translating visual plans into robot actions. This&That uses language-gesture conditioning to generate video predictions, as a succinct and unambiguous alternative to existing language-only methods, especially in complex and uncertain environments. These video predictions are then fed into a behavior cloning architecture dubbed Diffusion Video to Action (DiVA), which outperforms prior state-of-the-art behavior cloning and video-based planning methods by substantial margins.