CrashML Analyzer: A Data-Driven Approach for Investigation and Fault Assignment in Autonomous Driving Collisions

This study investigates factors contributing to autonomous vehicle (AV) accidents and proposes an automated fault determination framework. A total of 563 accident reports from the State of California Department of Motor Vehicles spanning from 2019 to 2024 were analyzed by converting unstructured standardized reports into structured data using custom extraction tools. Analysis of these reports reveal that AVs were not at fault in 69.4% of cases, and were fully at fault for 22.6% of the cases. The proposed method uses these reports to provide an early indicator of fault likelihood and potentially replaces tedious manual review. Machine Learning (ML) and Natural Language Processing techniques were used to replicate the reported faults, achieving 96% average accuracy across three models: Gradient Boosting, Linear Regression, and Random Forest. Through feature engineering techniques in semantic feature extraction from narrative accident descriptions, quantifiable variables were obtained and aided a robust fault classification performance across diverse collision scenarios with full cross-validation testing. Key contributing factors included the impact location (damage area), vehicle movement, and environmental conditions (weather, road, and lighting conditions). An open-source web-based system is also provided to demonstrate real-time accident report uploads and automated fault analysis, as well as a comprehensive system with source code, enabling scalable analysis of larger datasets for a reproducible, data-driven framework for a full assessment. The framework enables attendees to use existing connected vehicle datasets to provide automated analysis with visualization revealing collision patterns and liability trends for safe AI system validation. This supports early assignment of responsibility in autonomous vehicle collisions as AV deployment accelerates globally; determining fault attributes is essential for establishing public trust, legal precedents, insurance frameworks, and policies.