Problem
Research questions and friction points this paper is trying to address.
Vision-Language-Action
spatial reasoning
geometric structure
GaussVLA: Geometry-Aware Spatial Reasoning for Vision-Language-Action Model
🤖 AI Summary
为解决VLA模型空间推理能力有限的问题，GaussVLA通过引入Gaussian Spatial Tokenizer和Depth-Aware Chain-of-Thought模块增强几何感知能力。
📝 Abstract
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.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Gaussian Spatial Tokenizer
Depth-Aware Chain-of-Thought
structured spatial reasoning
🔎 Similar Papers
No similar papers found.
