VALAR: An AI-driven approach to transforming crash reporting for law enforcement agencies

Accurate and efficient crash reporting is a cornerstone of transportation safety, yet current methods suffer from inconsistencies, time-consuming manual data entry, and officer safety concerns at crash scenes. This research presents the Vehicle Accident Lens for Automated Reporting (VALAR), an AI-powered system designed to modernize police crash reporting by leveraging multimodal large language models, computer vision, and agentic AI. VALAR enhances crash data collection by enabling officers to capture reports through conversational interfaces, voice dictation, and automated image processing rather than traditional form-based input. The system integrates with third-party data sources, including geolocation, weather, and traffic conditions, to enrich reports in real time. AI-driven automation improves data accuracy while reducing the administrative burden on officers, allowing them to focus on public safety. VALAR adheres to national crash data standards, including MMUCC and MIRE, ensuring interoperability with existing state and federal reporting systems. This Phase I research establishes VALAR’s feasibility through user-centered design, prototype development, and pilot testing with law enforcement agencies. The project’s findings will inform scalable, production-level implementation in Phase II, ultimately contributing to improved traffic safety, streamlined law enforcement workflows, and enhanced data-driven decision-making in transportation management.