Advanced driver assistance systems (ADASs) assist drivers by alerting them of the occurrence of events based on the sensing capabilities of the vehicle, reducing the effort required by drivers. Most vehicles that are recently launched vehicles have been endowed with ADAS, thereby significantly reducing traffic accidents. However, the Insurance Institute for Highway Safety has reported that traffic accidents caused by driver negligence may increase as drivers have become accustomed to using ADAS. Therefore, drivers must be provided with sufficient information on the appropriate use of ADAS through user manuals. In this study, the regulations regarding the presentation of the operational design domain (ODD) in ADAS user manuals were analyzed. The results indicated that most user manuals do not sufficiently specify the ODD, which is claimed important by various organizations for ensuring safe driving. Additionally, the expression of the limitations and performance of ADAS is ambiguous because most countries are not regulated to explicitly present the ODD when writing ADAS user manuals. Therefore, in this study, the ODD guidelines for presenting ADAS specific to vehicle manufacturers and governments have been outlined in addition to guidelines for drivers on using ADAS. These guidelines can contribute to the development of clear ADAS user manuals, which in turn can ensure the safe driving of ADAS-equipped vehicles.

This study aims to identify the factors that influence the occurrence of traffic accidents to improve motorway traffic safety. Various data, including the frequency of traffic accidents, traffic volume, geometric structure, and congestion level, were collected from individual sections of motorways in South Korea. Using the collected data, a traffic accident frequency prediction model was developed by applying an explainable artificial intelligence (AI)-based approach. The developed deep neural network model was combined with Shapley Additive Explanations to identify the variables that significantly affect the frequency of traffic accidents. The analysis identified five significant factors: segment length, total traffic volume, the proportion of truck traffic, the number of dangerous driving behaviors, and the duration of congestion. The results demonstrate the potential of using explainable AI in predicting traffic accident frequency. By identifying the factors that influence traffic accidents using this model, we can pinpoint areas for improvement, which may ultimately help reduce highway traffic accidents.

The data storage system for automated driving (DSSAD) records how an automated driving system (ADS) behaves, but current requirements leave important gaps. UN Regulation No. 157 requires only timestamped state changes, and the time-series recording recently introduced by UN Regulation No. 185 starts only when a collision is detected or the event data recorder (EDR) is triggered and covers just 7 s around the trigger. ADS behavior in near misses and in minor contacts that are not detected as collisions may therefore still go unrecorded, and this window is much shorter than the 20 s pre-crash period now required for EDRs. This study proposes an evidence-based, two-layer DSSAD recording scheme. The timestamp layer covers the entire ADS, following the event classification of the Validation Method for Automated Driving, and designates seven hazard-related events, including non-collision events, as time-series triggers. The time-series layer records perception, decision, and control data at 10 Hz or higher from 20 s before to 10 s after a trigger. The scheme was derived from eight incident investigations in South Korea and the United States, a binary logistic regression screening of 842 National Highway Traffic Safety Administration crash reports, and two rounds of review by 13 incident investigation practitioners. It allows ADS behavior to be reconstructed in hazardous events that end without a collision and over a period long enough to cover the events leading up to an incident.