iFlyBot-VLA Technical Report

TL;DR

iFlyBot-VLA introduces a large-scale vision-language-action model trained with a novel dual-level action framework, combining latent and structured actions, to improve robotic manipulation understanding and performance.

Contribution

The paper presents a new VLA model with a dual-level action representation and a mixed training strategy, enhancing perception and reasoning in robotic manipulation tasks.

Findings

Outperforms existing methods on LIBERO Franka benchmark

Achieves high success rates in real-world manipulation tasks

Effectively integrates latent and structured actions for better control

Abstract

We introduce iFlyBot-VLA, a large-scale Vision-Language-Action (VLA) model trained under a novel framework. The main contributions are listed as follows: (1) a latent action model thoroughly trained on large-scale human and robotic manipulation videos; (2) a dual-level action representation framework that jointly supervises both the Vision-Language Model (VLM) and the action expert during training; (3) a mixed training strategy that combines robot trajectory data with general QA and spatial QA datasets, effectively enhancing the 3D perceptual and reasoning capabilities of the VLM backbone. Specifically, the VLM is trained to predict two complementary forms of actions: latent actions, derived from our latent action model pretrained on cross-embodiment manipulation data, which capture implicit high-level intentions; and structured discrete action tokens, obtained through frequency-domain transformations of continuous control signals, which encode explicit low-level dynamics. This dual supervision aligns the representation spaces of language, vision, and action, enabling the VLM to directly contribute to action generation. Experimental results on the LIBERO Franka benchmark demonstrate the superiority of our frame-work, while real-world evaluations further show that iFlyBot-VLA achieves competitive success rates across diverse and challenging manipulation tasks. Furthermore, we plan to open-source a portion of our self-constructed dataset to support future research in the community