As autonomous driving technology continues to advance, there is a growing need to transition from fixed routes to user-defined destinations, allowing passengers to set their preferred routes for autonomous vehicles. In response to this demand, we have developed a robust human-machine interface (HMI) that leverages both in-vehicle and external server resources to facilitate seamless communication between users and autonomous driving systems. The conventional approach to autonomous driving predominantly relied on pre-programmed routes and maps, limiting the flexibility for passengers to select their desired destinations. In contrast, our research endeavors to empower passengers with the ability to define their route preferences using a tablet-based interface, thereby enhancing the user experience and making autonomous driving more intuitive and accommodating. This paper presents the technical details of our HMI architecture, including the connection between autonomous driving system and servers, and emphasizes the importance of security, data privacy, and real-time communication between the vehicle and servers.

Under the current Motor Vehicle Management Act, braking systems are managed as vehicle inspection standards and have been continuously implemented for traffic safety of driving. We would like to conduct an experiment to analyze the actual effectiveness of the braking system inspection, and analyze the effectiveness of the experiment to present the importance of the braking system inspection. We will analyze the experimental results before and after the failure by artificially creating a failure in the braking system using the actual vehicle. First of all, we present the process of selecting the braking system as the subject of an nonconformity experiment, and determine the method of reproducing the braking system failure. Next, on-site braking experiments are conducted by conditions such as braking start speed, braking force, failure conditions for each braking system direction (left wheel or right wheel), and road surface conditions. Finally, we will conclude this study by organizing the test effectiveness (cost, risk, etc.) analysis based on the results of the experiment.

Development of the autonomous vehicle enables various sitting postures for the occupants. To assess the potential danger of certain sitting postures with respect to occupant safety and to find appropriate restraint systems for the occupants, MADYMO PSM (prescribed structural motion) models were developed for several sitting postures in several test modes like KNCAP side pole, AE-MDB, and USNCAP side pole test. Postures like normal, relaxed, and Long-slide relaxed postures in side impacts were studied. Effectiveness of current restraint systems in relaxed posture were assessed. Current side impact restraints are designed for normal posture which does not optimally provide protection for various occupant postures considered for autonomous vehicles.

Autonomous vehicles use sensors like LiDAR, radar, cameras, and ultrasonic sensors to collect real-time environmental data. Efficient cleaning methods for these sensors are essential to counter performance degradation from contaminants like snow, dust, sand, and mud. This study examined the impact of washer pressure, spray duration, spray angle, and target point on optimizing LiDAR sensor cleaning. Experimental design data revealed that washer pressure and spray duration are key factors. Due to physical limits and spray pump costs, increasing washer pressure infinitely is impractical. Optimal spray pressure was determined to be 7-8 bar. Spray duration affects water usage, so longer times mean higher consumption, necessitating consideration of the washer's tank capacity. The study suggests repeated short-duration sprays. The best cleaning efficiency occurred with a target point about 10 mm away and a spray angle of 20-30°. Using the central composite design of response surface methodology minimized experiments and helped select optimization factors.

The growing demand for traffic accident analysis requires accurate and cost-effective traffic accident scene modeling technology. The purpose of this study is to analyze which technologies can accurately map the real-world environment while considering time and cost. The methodology is to calculate the error of the maps produced by other methods based on the precision maps produced by multi-sensor fusion technology, which are considered to be highly accurate. While it is important to create maps with high accuracy, it is also important to use less time and money. Therefore, we analyze the time and cost of generating maps with each method. The results show which method can be recommended for modeling traffic accident scenes.

With the growing importance of Event Data Recorder (EDR) data in accident analysis, this study developed a program to efficiently analyze and verify the reliability of EDR data. EDR data, typically stored in PDF format, presents challenges for extraction and analysis. To overcome these issues, we implemented functions to convert PDF-based EDR data into Excel, visualize key information, and automatically generate files compatible with PC-Crash for accident reconstruction. In addition, we structured the accident reconstruction process and analyzed real accident cases focusing on vehicle speed and Electronic Stability Control (ESC) activation. The results show that the use of kinetic models and vehicle speed data enhances the accuracy of reconstructions, particularly in high-speed driving and ESC activation scenarios. By addressing these challenges, this study aims to improve the accuracy of accident analysis using EDR data and provide a structured methodology for accident reconstruction.

Tower crane accidents are one of the major safety concerns in construction sites. Conventional accident analyses have faced challenges in accurately identifying the root causes due to a lack of objective data. To address this issue, this study developed a tower crane work record device and evaluated its performance through field application. The developed device is capable of real-time data collection and storage of various work-related data, including load, working radius, slewing angle, and wind speed. By installing the device in actual field sites and collecting data over an extended period, the stability of the system and the accuracy of the data were verified.

This study examined overseas legislative trends for the deployment of Lv.4 AV, analyzed the main contents and meaning of the performance certification and conformity approval system in the 「Automated Driving Vehicles Act」, and looked at some issues that need to be improved. Although the performance certification and conformity approval system has limitations on subject, object, and operating scope, it can be seen as an advancement in that it provides a legal basis for the deployment of Lv.4 AV, unlike temporary operation or operation within a pilot zone. In addition, it has the meaning of a compromise system that can offset unconstitutional factors due to administrative legislative inaction and promote harmony with international standards by temporarily supplementing the self-certification system until safety standards are established according to the 「Motor Vehicle Management Act」, thereby allowing automobile manufacturers, etc. to deploy Lv.4 AV.

As Level 4 autonomous vehicles approach widespread adoption, external Human-Machine Interfaces (eHMI) have been proposed as effective mediators to replace the communication role of human drivers. This study investigates pedestrians' cognitive reaction times to eHMI-equipped vehicles and proposes a structured information delivery method for effective communication between autonomous vehicles and pedestrians. A virtual reality (VR) experiment with 57 participants was conducted, where participants observed an LED-equipped vehicle decelerating from 35 km/h to a full stop over 50 meters in 10.62 seconds. Participants recorded their reaction times at three key moments: recognizing the vehicle, identifying it as autonomous, and judging its intention to stop. Based on these results, a three-stage information delivery method was developed. Stage 1 introduces vehicle control status at 30 meters to help pedestrians "recognize" it as autonomous. Stage 2 provides acceleration or deceleration cues at 15 meters to allow pedestrians to "judge" its movement. Stage 3 conveys intent to yield or stop at 11 meters, enabling pedestrians to "respond" confidently. This stepwise approach ensures continuous assessment of the vehicle's state and intentions, affording pedestrians sufficient time to make informed decisions and enhancing their confidence in road-crossing scenarios. This study contributes to the design of eHMI by incorporating actual pedestrian cognitive reaction times, offering valuable insights for improving road safety in the era of autonomous vehicles and promoting pedestrian safety.

This study aims to evaluate and monitor the overall performance of autonomous driving systems (ADS) and help DSSAD investigate accidents. In the current DSSAD system, only ADS status values are recorded with a timestamp method, and data retrieval methods and data reporting formats considering real accident cases are not reflected, so DSSAD may difficult to identify the cause of the accident quickly. Therefore, current DSSAD-related regulations and risk situations during autonomous driving in Korea were considered. This presented improvements to the DSSAD data management plan to contribute to ADS incident investigation.

This study aimed to enhance the safety of school buses by investigating user satisfaction and analyzing the prioritization of vehicle safety devices using the KANO model and Timko's customer satisfaction coefficients. The results of the user experience-based satisfaction survey indicated a high overall satisfaction level, with particularly positive evaluations of safety-related devices. However, some respondents expressed concerns regarding the narrowness of seat specifications and legroom. To examine the relationship between satisfaction scores and demographic characteristics (gender, grade, and Use of existing school bus), random forest and chi-squared analyses were conducted, revealing significant factors affecting satisfaction scores. Subsequently, a KANO model analysis was performed to derive the prioritization of additional vehicle devices for improving the safety of school buses. Among the eight vehicle safety devices analyzed through the KANO model, some were categorized as One-dimensional quality, while others were classified as Attractive quality. The prioritization of safety devices was further established using Timko's customer satisfaction coefficients. The findings of this study are expected to contribute valuable insights for the commercialization of technologies aimed at improving the safety of school buses in the future.

Autonomous driving technology nowadays targets to level 4 or beyond, but the researchers are faced with some limitations for developing reliable driving algorithms in diverse challenges. To promote the autonomous vehicles to spread widely, it is important to properly deal with the safety issues on this technology. Among various safety concerns, the sensor blockage problem by severe weather conditions can be one of the most frequent threats for lane de-tection algorithms during autonomous driving. To handle this problem, the importance of the generation of proper datasets is becoming more significant. In this paper, a synthetic lane dataset with sensor blockage is suggested in the format of lane detection evaluation. Rain streaks for each frame were made by an experimentally established equation. Using this dataset, the degradation of the diverse lane detection methods has been verified. The trend of the per-formance degradation of deep neural network-based lane detection methods has been analyzed in depth. Finally, the limitation and the future directions of the network-based methods were presented.