Achieving generalization in robotic manipulation remains a critical challenge, particularly for unseen scenarios and novel tasks. Current Vision-Language-Action (VLA) models, while building on top of general Vision-Language Models (VLMs), still fall short of achieving robust zero-shot performance due to the scarcity and heterogeneity prevalent in embodied datasets. To address these limitations, we propose FSD (From Seeing to Doing), a novel vision-language model that generates intermediate representations through spatial relationship reasoning, providing fine-grained guidance for robotic manipulation. Our approach combines a hierarchical data pipeline for training with a self-consistency mechanism that aligns spatial coordinates with visual signals. Through extensive experiments, we comprehensively validated FSD's capabilities in both "seeing" and "doing," achieving outstanding performance across 8 benchmarks for general spatial reasoning and embodied reference abilities, as well as on our proposed more challenging benchmark VABench. We also verified zero-shot capabilities in robot manipulation, demonstrating significant performance improvements over baseline methods in both SimplerEnv and real robot settings. Experimental results show that FSD achieves 40.6% success rate in SimplerEnv and 72% success rate across 8 real-world tasks, outperforming the strongest baseline by 30%.

翻译：实现机器人操作的泛化能力仍是一项关键挑战，尤其是在应对未见场景和新颖任务时。当前的视觉-语言-动作（VLA）模型虽基于通用视觉-语言模型（VLM）构建，但由于具身数据集的稀缺性和异质性，仍难以实现鲁棒的零样本性能。为突破这些局限，我们提出FSD（从感知到行动）——一种通过空间关系推理生成中间表征的新型视觉-语言模型，可为机器人操作提供细粒度引导。该方法结合了用于训练的分层数据流水线，以及将空间坐标与视觉信号对齐的自一致性机制。通过广泛实验，我们全面验证了FSD在"感知"与"行动"两方面的能力：在8个通用空间推理与具身参考能力基准测试，以及我们提出的更具挑战性的VABench基准上均取得卓越性能。此外，我们验证了机器人在零样本操作中的能力，在SimplerEnv和真实机器人环境中均展现出相比基线方法的显著性能提升。实验结果表明，FSD在SimplerEnv中达到40.6%的成功率，在8项真实世界任务中达到72%的成功率，较最强基线方法提升30%。