arXiv cs.AI· Wanchun Ni, Tao Qi, Leonel Aguilar, Jiugeng Sun, Marlene Wagner, Verena Zimmermann, Mennatallah El-Assady·· 4 小时前AI 评分37
STRIDE：跨多场景文本到行人轨迹对齐的自动化评估框架
STRIDE: Automated Evaluation of Text-to-Trajectory Alignment across Diverse Contexts
AI 导读
STRIDE 是首个评估场景描述与行人轨迹上下文对齐的自动化框架，无需人工轨迹数据即可实现可验证、可扩展的评估。其基准 STRIDE-Bench 包含 1K 场景、6K 行为问题和 11K 测量项，覆盖 30 张真实地图，人工一致性达 80%。对多个文本到轨迹模型的评估显示，其上下文对齐能力有限，细粒度条件控制仍是难题。
正文 · 原文
Abstract:Language-conditioned trajectory generation is here, but its evaluation has not kept pace. Existing pedestrian trajectory metrics compare trajectories with real-world human data. This does not scale to text-to-trajectory generation across diverse contexts, as collecting human trajectories for every scenario is costly and infeasible. Moreover, pedestrian behavior is heterogeneous and context-dependent, with no single metric as the correct answer, and current evaluation frameworks are not transferable to this domain. These challenges make scalable, reliable evaluation difficult. We introduce STRIDE, the first framework for evaluating context alignment between scenario descriptions and pedestrian trajectories. STRIDE addresses these challenges through three design choices. First, we derive our VRDST evaluation protocol from sociological theories to define a complete evaluation space. Second, it decomposes high-level context into scenario-adaptive behavioral questions. Third, every question is resolved against a deterministic measurement tool library that yields reproducible answers. Together, STRIDE enables verifiable, automated, and scalable evaluation across diverse contexts without requiring human trajectory data. We instantiate STRIDE in the crowd domain as STRIDE-Bench, comprising 1K scenarios, 6K behavioral questions, and 11K measurements with calibrated expected answers across 30 real-world maps. Comprehensive human validations show that STRIDE-Bench is consistent with human behavior and judgment, achieving 80% human agreement. We further evaluate several text-to-trajectory models, finding limited context-alignment capability and persistent challenges in fine-grained context conditioning. We believe that the STRIDE framework provides a first step toward principled evaluation of context-aligned pedestrian trajectory generation.
| Comments: | Accepted at NeurIPS 2026, Evaluations & Datasets Track |
| Subjects: | Artificial Intelligence (cs.AI); Computers and Society (cs.CY) |
| Cite as: | arXiv:2609.34799 [cs.AI] |
| (or arXiv:2609.34799v3 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2609.34799 arXiv-issued DOI via DataCite |
Submission history
From: Wanchun Ni [view email]
[v1]
Mon, 28 Sep 2026 10:01:45 UTC (2,661 KB)
[v2]
Tue, 29 Sep 2026 09:09:24 UTC (2,661 KB)
[v3]
Wed, 4 Nov 2026 14:44:10 UTC (2,664 KB)
来源：arXiv cs.AI · arxiv.org