Hy-Embodied-0.5-VLA: From Vision-Language-Action Models to a Real-World Robot Learning Stack

2026-06-12

AI Summary

This work proposes and implements an end-to-end robotic learning system that establishes a complete closed-loop pipeline from multimodal perception to action execution in the real world. Addressing the common disconnect between algorithmic development and physical deployment in existing approaches, the study presents the first deep integration of a vision–language–action (VLA) foundation model into a full robotic learning engineering stack. This stack encompasses data collection, continual pretraining, supervised fine-tuning, reinforcement learning-based post-training, and real-world deployment. The resulting system demonstrates strong generalization and robustness on complex tasks, effectively bridging the gap between algorithmic prototypes and their efficient realization on physical robots.

Abstract

In this report, we present Hy-Embodied-0.5-VLA, abbreviated as HyVLA-0.5, an end-to-end system that spans the full robot learning stack: data collection, model design, continued pre-training and supervised fine-tuning, RL post-training, and real-world deployment. Each component serves a distinct role in this stack.

Problem

• Vision-Language-Action Models
• Robot Learning
• End-to-End System
• Real-World Deployment
• Embodied AI

Innovation

• Vision-Language-Action
• robot learning stack
• end-to-end system
• real-world deployment
• continued pre-training