用3D高斯表示视觉空间结构，提升机器人动作预测的几何推理能力。
GaussVLA: Geometry-Aware Spatial Reasoning for Vision-Language-Action Model
- 将视觉与深度特征转化为紧凑的3D高斯令牌，保留几何结构信息。
- 在LIBERO上达93.5%平均成功率，空间任务全成功，仅用200M参数。
- 适合需要高效精准空间推理的机器人控制场景。
视觉-语言-动作（VLA）模型将视觉观测编码为无内在几何结构的2D图像块，而稠密单目深度仅注入每像素标量值，无法表达表面朝向或几何置信度，导致策略缺乏结构化空间推理能力。本文提出GaussVLA，一种基于Mamba的VLA模型，引入两个自定义模块：高斯空间分词器（GST）将冻结的语义与深度特征升维为紧凑的3D高斯令牌，并通过学习查询聚合几何显著区域；以及深度感知思维链（DA-CoT），在语言和时序流条件下进行结构化、非自回归的几何推理。在仿真与真实世界评估中，GaussVLA展现出强大的空间操作性能，同时保持参数高效。在LIBERO数据集上，平均成功率达93.5%，空间任务套件成功率为100.0%，仅使用200M参数，相比SpatialVLA相对平均成功率提升19.7%，且参数更少。
原文摘要 · Abstract (English)
Vision-Language-Action (VLA) models encode visual observations as flat 2D patch tokens that carry no intrinsic geometric structure, and augmenting them with dense monocular depth injects per-pixel scalar values that encode neither surface orientation nor geometric confidence. This leaves the policy with limited structured spatial reasoning for action prediction. We propose GaussVLA, a Mamba-based VLA that incorporates two custom modules: Gaussian Spatial Tokenizer (GST) to lift frozen semantic and depth features into compact 3D Gaussian tokens, pools geometrically salient regions with learned queries, and \emph{Depth-Aware Chain-of-Thought (DA-CoT)} that performs structured, non-autoregressive geometric reasoning under language and flow-time conditioning. Across both simulation and real-world evaluations, GaussVLA demonstrates strong spatial-manipulation performance while remaining parameter-efficient. On LIBERO, it achieves 93.5% average success and 100.0% success on the Spatial suite with only 200M parameters, improving over SpatialVLA by 19.7% relative average success while remaining significantly more parameter-efficient.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。