Hy-Embodied-0.5-VLA: From Vision-Language-Action Models to a Real-World Robot Learning Stack
He Zhang, Lingzhu Xiang, Haitao Lin, Zeyu Huang, Minghui Wang, Dingyan Zhong, Yubo Dong, Yihao Wu, Yongming Rao, Dongsheng Zhang, Wanjia He, Ling Chen
arXiv:2606.14409v112.21 citations
Predicted impact top 28% in RO · last 90 daysOriginality Synthesis-oriented
AI Analysis
This work provides a complete, reproducible pipeline for embodied AI, but the components are largely incremental integrations of existing methods.
Hy-Embodied-0.5-VLA (HyVLA-0.5) is an end-to-end robot learning system covering data collection, model design, pre-training, fine-tuning, RL post-training, and deployment. The system achieves a 78% success rate on real-world manipulation tasks, outperforming prior baselines by 15%.
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.