Search papers, labs, and topics across Lattice.
This paper systematically investigates the design space of Vision-Language-Action (VLA) models by re-examining foundational components, perception essentials, and action modeling perspectives under a unified framework. Through controlled experiments starting from an RT-2/OpenVLA-like baseline, the authors identify 12 key findings that contribute to building effective VLA models. The resulting model, VLANeXt, achieves state-of-the-art performance on LIBERO and LIBERO-plus benchmarks and exhibits strong real-world generalization.
Forget ad-hoc VLA design: here are 12 key ingredients, validated in a unified framework, for building performant Vision-Language-Action models.
Following the rise of large foundation models, Vision-Language-Action models (VLAs) emerged, leveraging strong visual and language understanding for general-purpose policy learning. Yet, the current VLA landscape remains fragmented and exploratory. Although many groups have proposed their own VLA models, inconsistencies in training protocols and evaluation settings make it difficult to identify which design choices truly matter. To bring structure to this evolving space, we reexamine the VLA design space under a unified framework and evaluation setup. Starting from a simple VLA baseline similar to RT-2 and OpenVLA, we systematically dissect design choices along three dimensions: foundational components, perception essentials, and action modelling perspectives. From this study, we distill 12 key findings that together form a practical recipe for building strong VLA models. The outcome of this exploration is a simple yet effective model, VLANeXt. VLANeXt outperforms prior state-of-the-art methods on the LIBERO and LIBERO-plus benchmarks and demonstrates strong generalization in real-world experiments. We will release a unified, easy-to-use codebase that serves as a common platform for the community to reproduce our findings, explore the design space, and build new VLA variants on top of a shared foundation.
