teLLMe System for Causal Analysis of Urban Driving Datasets
July 17, 2026
teLLMe enables causal inference from observational dashcam data by mapping natural language queries to structured causal queries via an LLM. The system uses the PC algorithm and DoWhy to provide effect estimates and summarized Causal Cards for urban traffic scenarios.
HOW THIS AFFECTS YOU
●
researcherYou can use this framework to move from observational correlations to causal estimates in video-derived datasets.