pith. sign in

arxiv: 2605.31148 · v1 · pith:L3QDQ6AUnew · submitted 2026-05-29 · 💻 cs.CV · cs.AI· cs.CL

SpatialAct: Probing Spatial Reasoning-to-Action Capabilities of VLM Agents in 3D Scenes

classification 💻 cs.CV cs.AIcs.CL
keywords spatialreasoningactionsmulti-turntasksagentscoherentcurrent
0
0 comments X
read the original abstract

Humans can effortlessly perceive spatial layouts, form cognitive representations, reason about spatial relations, and translate such reasoning into actions in everyday 3D environments. Although recent vision-language models (VLMs) have shown promising performance on observation-conditioned spatial perception and reasoning tasks, it remains unclear whether they can build coherent spatial understanding, act upon it, and refine their actions through multi-turn feedback. To study this problem, we introduce \textbf{SpatialAct}, a simulator-grounded benchmark for probing \textit{action-conditioned spatial reasoning} in 3D scenes. Starting from the most challenging setting, Multi-turn Interactive Refinement, we further design its decomposed counterpart, Single-step Error Detection and Fix, together with five fundamental spatial ability tasks to diagnose the underlying causes of model failures. Experiments reveal a clear reasoning-to-action gap: current VLMs can perform well on isolated spatial reasoning tasks, but struggle to maintain coherent spatial beliefs and produce reliable actions during multi-turn feedback, substantially underperforming humans. These results suggest that current VLM agents still lack robust spatial state tracking under action-induced environment changes, even when low-level control is abstracted away.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.