*2.5. Summary*

In summary, diverse studies had been conducted about the tra ffic impacts of AVs. Most studies verified that when AVs are introduced, LOS, speed, and road volume (capacity) are improved and tra ffic accidents decreased even in a condition of mixed tra ffic with HVs. Also, cognition studies on AVs social acceptability were carried out. In addition, AI is actively being applied and developed in various ways for AVs. There have been various studies such as AVs control study based on CNNs, End-to-End, tra ffic volume forecasting based on DNN-Based Tra ffic Flow prediction (DNN-BTF), and short tra ffic flow prediction using LSTM.

Although studies about AVs introduction and acceptance surveys have been actively delved into, it is found that there exist limitations in building the foundation for commercialization. In addition, researches on legal and ethical issues on AVs tra ffic accidents have been carries out from various perspectives, but opinions have been continuously raised whether it is appropriate to answer all the ethical issues related with AVs tra ffic accidents as pre-conditions of the commercialization. Similarly, most AVs technology researches have gradually advanced to study related to the technology supplementation of themselves, but the researches related with the prevention of AVs tra ffic accidents have been found insu fficient. Therefore, as proposed in the German and U.S. ethics guidelines and NVIDIA report [47], this study aims to lay the foundation for the design of the PADS using AI as a way to prevent AVs tra ffic accidents in advance.
