2026 ACECNJ Fall Conference

Artificial intelligence (AI) and machine learning (ML) are transforming traffic safety analysis by enabling faster, more scalable, and data-driven insights that extend beyond traditional crash-based approaches. In this case study, a New Jersey urban community explored how AI-powered analysis of near-miss events, instances where collisions were narrowly avoided, could supplement conventional safety evaluations in support of the City’s Vision Zero initiative and Complete Streets goals. Unlike crash data, which typically informs reactive improvements after incidents occur, machine learning-driven near-miss analytics provide a proactive lens by detecting patterns in frequent, high-risk interactions and identifying emerging hazards before crashes take place. Support was provided for the collection and evaluation of near-miss data from multiple vendors, leveraging AI and ML techniques to compare detection methodologies, data outputs, and analytical approaches. This evaluation identified strengths, limitations, and opportunities for application across platforms. The effort was conducted in parallel with multimodal safety planning initiatives, where AI-enhanced insights informed concept development focused on improving safety for pedestrians, cyclists, transit users, and motorists. The case study assessed how AI- and ML-enabled near-miss analytics could enhance routine traffic engineering practices, including warrant analyses, safety studies, and crash investigations, by providing a broader, more dynamic, and near real-time understanding of roadway risk. Findings highlighted the value of near-miss analytics in identifying conflict patterns at scale, while also identifying challenges related to data consistency, model transparency, interpretation, and integration into established workflows.