TreasureX Researchcs.AIteLLMe: A Tool to Explore Cause-and-Effect in Traffic Data
Qiwei Li, Jorge Ortiz · July 16, 2026
Traffic cameras generate heaps of video data, but figuring out cause and effect—like whether rain actually makes traffic denser—is tricky because we can't run controlled experiments on real roads. Researchers introduce teLLMe, a system that helps analysts explore causal relationships from observational dashcam data. It starts by structuring raw annotations into events, then uses a causal learning algorithm (PC algorithm) with stability checks to propose possible cause-effect links. Users can ask natural-language questions (e.g., "Does rain increase density during peak hours?"), which the system translates into structured queries. teLLMe returns a "Causal Card" summarizing estimated effects, adjustment variables, and assumptions, plus a plain-English explanation. Case studies on traffic events show it can surface plausible relationships like weather affecting density, while making uncertainties clear. The system is designed for hypothesis generation, not definitive answers—a tool for experts to reason with data. This approach could help city planners and traffic engineers test ideas without costly field experiments, potentially leading to smarter traffic management.
Read the full paper on arXiv: https://arxiv.org/abs/2607.15254
Educational content only, generated with AI assistance from public research abstracts. Not investment advice. Thank you to arXiv for use of its open access interoperability.