Universal Pose Pretraining for Generalizable Vision-Language-Action Policies
This addresses the challenge of generalizable robot policies for vision-language-action tasks, offering an incremental improvement over existing methods by decoupling perception and action training.
The paper tackles the problem of feature collapse and low training efficiency in Vision-Language-Action models by proposing Pose-VLA, a decoupled paradigm that separates training into pre-training for universal 3D spatial priors and post-training for embodiment alignment, achieving state-of-the-art results with a 79.5% average success rate on RoboTwin 2.0 and 96.0% on LIBERO.
Existing Vision-Language-Action (VLA) models often suffer from feature collapse and low training efficiency because they entangle high-level perception with sparse, embodiment-specific action supervision. Since these models typically rely on VLM backbones optimized for Visual Question Answering (VQA), they excel at semantic identification but often overlook subtle 3D state variations that dictate distinct action patterns. To resolve these misalignments, we propose Pose-VLA, a decoupled paradigm that separates VLA training into a pre-training phase for extracting universal 3D spatial priors in a unified camera-centric space, and a post-training phase for efficient embodiment alignment within robot-specific action space. By introducing discrete pose tokens as a universal representation, Pose-VLA seamlessly integrates spatial grounding from diverse 3D datasets with geometry-level trajectories from robotic demonstrations. Our framework follows a two-stage pre-training pipeline, establishing fundamental spatial grounding via poses followed by motion alignment through trajectory supervision. Extensive evaluations demonstrate that Pose-VLA achieves state-of-the-art results on RoboTwin 2.0 with a 79.5% average success rate and competitive performance on LIBERO at 96.0%. Real-world experiments further showcase robust generalization across diverse objects using only 100 demonstrations per task, validating the efficiency of our pre-training paradigm.