This paper attempts to solve the problem of identifying the critical factors of road design on human driver (HD) accidents and automated vehicle (AV) failures. This study is of immediate importance for the safety of the road, which must facilitate the wide emergence of commercialized vehicle automation technologies. We overcome the weakness of the current literature that relies on simulation to capture the safety of the road for AVs, by using real-world field data. We drive commercially available vehicles with driving automation technologies and collect data on their driving failures on the real road network of South Korea over a span of 6,730 kilometers. We additionally gather data on detailed road geometry like road curvature and road width, traffic operation, and HD accidents for the same roads. We implement a machine learning method to predict if a given road link has had any HD accidents and AV failures. A cross-validation experiment shows high prediction accuracy for HD accidents around 94% and for AV accidents around 82%, despite the unbalanced dataset on safe and unsafe road links. Moreover, the machine learning results are interpreted with Shapley Additive exPlanation method, and we identify critical safety factors of road geometry and operation that are common and distinct for the HDs and AVs. We find that the safety of human-driven and AVs are both affected by the traffic operation variables, such as the traffic speed and volume during peak and non-peak hours. However, AV failures are affected by road geometry, such as road curvature, road width, and spacing between streetlights, more than the HD accidents. AVs are safer in road links with higher speed limits than HDs. Also, AV failures are affected differently by structures, such as shock absorbers and junctions, to the HD accidents. From these findings, we provide policy recommendations on the safe road design for both HDs and AVs.

In addition to addressing the labor shortage due to an aging population, the transition to autonomous vehicle (AV)-based mobility services offers enhanced efficiency and operational flexibility for public transportation. However, much of the existing focus has been on improving AV safety without fully considering road conditions and real-world service demand. This study contributes to the literature by proposing a comprehensive framework for efficiently integrating AV-based mobility services at the network level, addressing these gaps. The framework analyzes and optimizes service networks by incorporating actual demand patterns, quantifying road segment difficulty from an AV perspective, and developing an optimization model based on these factors. The framework begins by quantifying the operational difficulty of road segments through an evaluation of Operational Design Domains (ODDs), providing a precise measure of AV suitability under varying road conditions. It then introduces a quantitative metric to assess operational feasibility, considering factors such as the service margin, costs, and safety risks. Using these metrics alongside Genetic Algorithms (GAs), the framework identifies an optimal service network that balances safety, efficiency, and profitability. By analyzing real-world data from different mobility services, such as taxis, Demand-Responsive Transport (DRT), and Special Transportation Services (STSs), this study highlights the need for service-specific strategies to optimize AV deployment. The findings show that optimal networks vary with demand patterns and road difficulty, demonstrating the importance of tailored network designs. This research provides a scalable, data-driven approach for integrating AV services into public transportation systems and lays the foundation for further improvements by incorporating dynamic factors and broader urban contexts.