Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

See, Remember, Explore: A Benchmark and Baselines for Streaming Spatial Reasoning

About

Spatial understanding is fundamental for embodied agents, yet most spatial VLMs and benchmarks remain offline-evaluating post-hoc QA over pre-recorded inputs and overlooking two crucial deployment-critical requirements: long-horizon streaming inference and active perception when the current view is insufficient. To address this gap, we introduce S3-Bench, a benchmark suite for streaming spatial question answering with active exploration, where queries are temporally grounded to specific timestamps and must be answered using only observations available up to that moment. S3-Bench adopts a dual-domain design, combining a scalable simulator with controllable trajectories and exploration actions, and real-world streaming videos that capture practical sensing artifacts for rigorous generalization evaluation. Overall, it spans 10K+ scenes and 26K+ trajectories, with dedicated training (S3-Train) and evaluation (S3-Eval) splits. We further propose AMF-VLM, which supports streaming spatial reasoning under bounded computing via (i) memory folding, which compresses long-horizon observations into compact structured memory, and (ii) active exploration, which outputs explicit actions (e.g. move/rotate/scan) to acquire missing evidence before answering. Extensive experiments demonstrate that, compared to models using identical training data, our approach yields improvements of 8.8% and 13.3% on the simulated and real splits of S3-Eval, respectively, while maintaining competitive transferability to standard spatial benchmarks.

Yuxi Wei, Wei Huang, Qirui Chen, Lu Hou, Xiaojuan Qi• 2026

Related benchmarks

TaskDatasetResultRank
Embodied Spatial ReasoningS3-Eval (real part)
Overall Score57.8
20
Streaming Spatial ReasoningS3-Eval Sim
Overall Accuracy62.9
20
Spatial UnderstandingVSIBench
Accuracy67.5
16
Spatial UnderstandingBLINK
Accuracy65.1
6
Spatial UnderstandingMMSI
Accuracy31.2
6
Active Spatial UnderstandingS3-Eval (simulation)
Overall Score62.9
4
Active Vision Spatial UnderstandingS3-Eval real
Overall Score57.8
4
Showing 7 of 7 rows

Other info

Follow for update