Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2512.16755 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918255110651904 |
|---|---|
| author | Wang, Siqi Liang, Chao Gao, Yunfan Yu, Erxin Li, Sen Li, Yushi Li, Jing Wang, Haofen |
| author_facet | Wang, Siqi Liang, Chao Gao, Yunfan Yu, Erxin Li, Sen Li, Yushi Li, Jing Wang, Haofen |
| contents | Vision-Language Models (VLMs) have made significant progress in explicit instruction-based navigation; however, their ability to interpret implicit human needs (e.g., "I am thirsty") in dynamic urban environments remains underexplored. This paper introduces CitySeeker, a novel benchmark designed to assess VLMs' spatial reasoning and decision-making capabilities for exploring embodied urban navigation to address implicit needs. CitySeeker includes 6,440 trajectories across 8 cities, capturing diverse visual characteristics and implicit needs in 7 goal-driven scenarios. Extensive experiments reveal that even top-performing models (e.g., Qwen2.5-VL-32B-Instruct) achieve only 21.1% task completion. We find key bottlenecks in error accumulation in long-horizon reasoning, inadequate spatial cognition, and deficient experiential recall. To further analyze them, we investigate a series of exploratory strategies-Backtracking Mechanisms, Enriching Spatial Cognition, and Memory-Based Retrieval (BCR), inspired by human cognitive mapping's emphasis on iterative observation-reasoning cycles and adaptive path optimization. Our analysis provides actionable insights for developing VLMs with robust spatial intelligence required for tackling "last-mile" navigation challenges. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_16755 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | CitySeeker: How Do VLMS Explore Embodied Urban Navigation With Implicit Human Needs? Wang, Siqi Liang, Chao Gao, Yunfan Yu, Erxin Li, Sen Li, Yushi Li, Jing Wang, Haofen Artificial Intelligence Vision-Language Models (VLMs) have made significant progress in explicit instruction-based navigation; however, their ability to interpret implicit human needs (e.g., "I am thirsty") in dynamic urban environments remains underexplored. This paper introduces CitySeeker, a novel benchmark designed to assess VLMs' spatial reasoning and decision-making capabilities for exploring embodied urban navigation to address implicit needs. CitySeeker includes 6,440 trajectories across 8 cities, capturing diverse visual characteristics and implicit needs in 7 goal-driven scenarios. Extensive experiments reveal that even top-performing models (e.g., Qwen2.5-VL-32B-Instruct) achieve only 21.1% task completion. We find key bottlenecks in error accumulation in long-horizon reasoning, inadequate spatial cognition, and deficient experiential recall. To further analyze them, we investigate a series of exploratory strategies-Backtracking Mechanisms, Enriching Spatial Cognition, and Memory-Based Retrieval (BCR), inspired by human cognitive mapping's emphasis on iterative observation-reasoning cycles and adaptive path optimization. Our analysis provides actionable insights for developing VLMs with robust spatial intelligence required for tackling "last-mile" navigation challenges. |
| title | CitySeeker: How Do VLMS Explore Embodied Urban Navigation With Implicit Human Needs? |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2512.16755 |