SOP: A Scalable Online Post-Training System for Vision-Language-Action Models

TL;DR

This paper presents SOP, a scalable online system for post-training vision-language-action models that enables real-time, multi-robot, multi-task learning directly in the physical environment, significantly improving performance and scalability.

Contribution

The paper introduces SOP, a novel online, distributed post-training framework for VLA models that integrates real-world interaction with scalable, multi-task learning in physical robots.

Findings

SOP improves VLA model performance across various tasks.

Post-training within hours of real-world interaction is feasible.

Performance scales near-linearly with the number of robots.

Abstract

Vision-language-action (VLA) models achieve strong generalization through large-scale pre-training, but real-world deployment requires expert-level task proficiency in addition to broad generality. Existing post-training approaches for VLA models are typically offline, single-robot, or task-specific, limiting effective on-policy adaptation and scalable learning from real-world interaction. We introduce a Scalable Online Post-training (SOP) system that enables online, distributed, multi-task post-training of generalist VLA models directly in the physical world. SOP tightly couples execution and learning through a closed-loop architecture in which a fleet of robots continuously streams on-policy experience and human intervention signals to a centralized cloud learner, and asynchronously receives updated policies. This design supports prompt on-policy correction, scales experience collection through parallel deployment, and preserves generality during adaptation. SOP is agnostic to the choice of post-training algorithm; we instantiate it with both interactive imitation learning (HG-DAgger) and reinforcement learning (RECAP). Across a range of real-world manipulation tasks including cloth folding, box assembly, and grocery restocking, we show that SOP substantially improves the performance of large pretrained VLA models while maintaining a single shared policy across tasks. Effective post-training can be achieved within hours of real-world interaction, and performance scales near-linearly with the number of robots in the fleet. These results suggest that tightly coupling online learning with fleet-scale deployment is instrumental to enabling efficient, reliable, and scalable post-training of generalist robot policies in the physical world.