VIMA: General Robot Manipulation with Multimodal Prompts

TL;DR

VIMA introduces a transformer-based model that uses multimodal prompts to unify various robot manipulation tasks, achieving strong zero-shot generalization and data efficiency in a new simulation benchmark.

Contribution

The paper presents a novel multimodal prompt framework and a scalable transformer-based robot agent, VIMA, for unified manipulation learning with systematic evaluation.

Findings

VIMA outperforms alternatives in zero-shot generalization by up to 2.9x success rate.

VIMA maintains high performance with 10x less training data, outperforming competitors.

A new benchmark with thousands of tasks and expert trajectories supports systematic evaluation.

Abstract

Prompt-based learning has emerged as a successful paradigm in natural language processing, where a single general-purpose language model can be instructed to perform any task specified by input prompts. Yet task specification in robotics comes in various forms, such as imitating one-shot demonstrations, following language instructions, and reaching visual goals. They are often considered different tasks and tackled by specialized models. We show that a wide spectrum of robot manipulation tasks can be expressed with multimodal prompts, interleaving textual and visual tokens. Accordingly, we develop a new simulation benchmark that consists of thousands of procedurally-generated tabletop tasks with multimodal prompts, 600K+ expert trajectories for imitation learning, and a four-level evaluation protocol for systematic generalization. We design a transformer-based robot agent, VIMA, that processes these prompts and outputs motor actions autoregressively. VIMA features a recipe that achieves strong model scalability and data efficiency. It outperforms alternative designs in the hardest zero-shot generalization setting by up to $2.9\times$ task success rate given the same training data. With $10\times$ less training data, VIMA still performs $2.7\times$ better than the best competing variant. Code and video demos are available at https://vimalabs.github.io/