CrossTraffic: An Open-Source Framework for Reproducible and Executable Transportation Analysis and Knowledge Management

Researchers Rei Tamaru and Bin Ran have introduced CrossTraffic, a groundbreaking open-source framework that addresses long-standing challenges in transportation engineering. The system treats transportation methodologies—like those in the Highway Capacity Manual (HCM)—as continuously deployable software infrastructure rather than static documents or proprietary tools. At its core is an executable computational engine for transportation analysis, accessible through standardized interfaces across platforms. Crucially, an ontology-driven knowledge graph encodes engineering rules and provenance, serving as a semantic validation layer that ensures analytical workflows follow proper procedures.

CrossTraffic's most innovative feature is its conversational interface that connects large language models (LLMs) to this validated execution environment through structured tool invocation. This allows users to perform complex transportation analyses using natural language while preventing procedurally invalid operations. Experimental results demonstrate remarkable performance: knowledge-graph-constrained execution achieves near-zero numerical error (MAE<0.50) across multiple LLMs and perfect detection of invalid analytical inputs in stress testing (F1~=1.0). The modular architecture supports integration of additional transportation manuals and research models, creating a foundation for an open, collaborative transportation science ecosystem with a reproducible computational core.