OmniJARVIS: Unified Vision-Language-Action Tokenization Enables Open-World Instruction Following Agents

TL;DR

OmniJARVIS introduces a unified multimodal tokenization approach for open-world instruction-following agents in Minecraft, enabling reasoning, planning, and acting through autoregressive transformers with enhanced decision-making capabilities.

Contribution

The paper proposes a novel unified tokenization method for multimodal interaction data, integrating behavior trajectories into pretrained language models for improved reasoning and decision-making in open-world environments.

Findings

Achieves strong performance on diverse Minecraft tasks

Effectively reasons, plans, and acts using unified tokens

Demonstrates scalable potential in multimodal interaction modeling

Abstract

This paper presents OmniJARVIS, a novel Vision-Language-Action (VLA) model for open-world instruction-following agents in Minecraft. Compared to prior works that either emit textual goals to separate controllers or produce the control command directly, OmniJARVIS seeks a different path to ensure both strong reasoning and efficient decision-making capabilities via unified tokenization of multimodal interaction data. First, we introduce a self-supervised approach to learn a behavior encoder that produces discretized tokens for behavior trajectories $\tau = \{o_0, a_0, \dots\}$ and an imitation learning policy decoder conditioned on these tokens. These additional behavior tokens will be augmented to the vocabulary of pretrained Multimodal Language Models. With this encoder, we then pack long-term multimodal interactions involving task instructions, memories, thoughts, observations, textual responses, behavior trajectories, etc into unified token sequences and model them with autoregressive transformers. Thanks to the semantically meaningful behavior tokens, the resulting VLA model, OmniJARVIS, can reason (by producing chain-of-thoughts), plan, answer questions, and act (by producing behavior tokens for the imitation learning policy decoder). OmniJARVIS demonstrates excellent performances on a comprehensive collection of atomic, programmatic, and open-ended tasks in open-world Minecraft. Our analysis further unveils the crucial design principles in interaction data formation, unified tokenization, and its scaling potentials. The dataset, models, and code will be released at https://craftjarvis.org/OmniJARVIS.