This paper presents an algorithm for setting dynamic LiDAR regions-of-interest (ROI) based on driving situations and paths to improve efficiency in autonomous vehicle perception using LiDAR data. The proposed algorithm selects only the relevant ROI for the vehicle movements, eliminating unnecessary computations and ensuring real-time, computationally efficient operations. Different driving situations, such as lane keeping, lane change, and passing intersections, are considered for dynamic ROI settings. In these situations, the algorithm considers the interested area based on the driving path and determines the requ ired perception range for safety, considering vehicle speed and acceleration values. To select LiDAR points more flexibly, the proposed algorithm uses a boundary-setting method that assigns indices to the LiDAR points instead of the conventional box-based ROI selection method. The algorithm has been validated using LiDAR data obtained from actu al driving tests conducted at the Sangam Au tonomous Driving Test Area. It has demonstrated improved efficiency by redu cing the total number of points and processing times.

This study is about some of the test conditions (scenarios) and results of various attempts to develop evaluation technology using SLED during the research and development of a crash safety test method considering the seating characteristics of passengers in autonomous vehicles. In previous studies of this study, several attempts were made to develop a sled-based 150-degree oblique vehicle-to-vehicle crash test method, but it was confirmed that there were limitations in reproducing a 150-degree oblique crash test with a sled. In order to develop 180 degrees 50% offset 56 km/h vehicle-to-vehicle crash test (MPDB) as a test mehtod using Sled, it is necessary to present a highly reproducible combination of Sled test pulse and rotation angle. To this end, an analysis study that could apply various conditions was conducted. When evaluating sled-based tests, we proposed a combination of sled test pulse and jig rotation angle that is expected to best reproduce the MPDB vehicle-to-Vehicle collision situation. In the future, we plan to confirm reproducibility through tests by applying the Sled test pulse and rotation angle combination proposed through this study in actual Sled tests.

The most important electronic devices installed in vehicles are V2X IVS, EDR, DTG, and NG e-Call IVS. Each of these devices serves a distinct purpose, they are collectively instrumental in enhancing the safety of vehicles. The objective of this study is to analyse the data collected or recorded by these devices in order to ascertain their usability of vehicle motion analysis such as vehicle traffic accidents. To this end, we undertook a review of the relevant standards, criteria and trends, conducted a comparative and analytical examination of the data collected or recorded by each device, and conducted the raw data collection and analysis using a real-vehicle. The analysis and experiments demonstrated that the analysis of complex data, rather than EDR data alone, is a suitable approach for traffic accident analysis in the context of C-ITS, SDV, and converged autonomous driving. Additionally, we can find the significance of data collected by V2X IVS for not only vehicle safety also vehicle motion analysis.

As autonomous driving technology advances, validation in virtual environments is becoming increasingly crucial, leading to ongoing efforts to reduce discrepancies between simulated and real-world vehicle dynamic behavior. Dynamic consistency is essential for evaluating the reliability and safety of autonomous algorithms. Although data-driven learning methods have been actively explored, simulation models cannot be fully replaced by learning approaches alone due to the importance of interaction with ground surfaces. Therefore, this study proposes a physics-based parameter estimation method for improving vehicle dynamics accuracy in virtual simulations for autonomous vehicles. The proposed methodology utilizes Physics-Informed Neural Networks (PINN) to effectively estimate parameters of both longitudinal and lateral vehicle dynamics models. This approach adheres to the physical laws governing vehicle dynamics and ensures robust interactions with the physics engines used in virtual environments, leveraging data-driven learning to enhance dynamic consistency. The proposed method is expected to serve as a core technology for verifying autonomous vehicle systems in virtual environments, contributing significantly to improving the safety and reliability of autonomous driving systems.

Considering the passenger's seating characteristics of autonomous vehicle passengers, we are conducting research on the development of test method using Sled that reflects crash safety tests for the development of automotive safety device test method. Previously, several attempts were made to develop a using Sled 150° inclined car-to-car crash test method, but it was confirmed that there were limitations in reproducing the car-to-car crash test in that mode. Next, we study to develop a sled test method that reflects a 180° 50% offset 56 km/h car-to-car crash test reflecting the reclined posture, and for this purpose, we studied the combination of test pulses and rotation angles expected to have high reproducibility through analysis research. As a result, a Sled-based test was conducted with the expected test pulse and jig rotation angle combination, and an appropriate Sled test method was studied through analysis of the results. A new concept of safety device was installed and the test method was conducted, and the appropriateness of the test method and the characteristics of human injury were also analyzed when the safety device was applied.

When converting internal combustion engine (ICE) vehicles to electric vehicles (EVs), the removal of the engine and the addition of high-capacity batteries shift the vehicle's center of gravity and alter the load distribution on the frame. For vehicles under 1 ton, these changes may lead to instability issues, such as rollover, understeer, or nose-up conditions, and localized stress on the frame, causing fatigue deformation or torsion. This study uses Fusion360 and Inventor Nastran to analyze structural stability under varying battery placements and proposes subframe designs to minimize deformation and enhance structural integrity.

As vehicles with autonomous driving systems continue to develop, future occupants eating positions may be more varied than today. It is possible that drivers may be able to operate vehicles in a reclined position, rather than the current standards eating position. Consequently, it is necessary to enhance the protection provided by occupant restraints, In this study we investigated the injury effects of occupants in a reclined position in a collision with the current level of vehicle restraint. The evaluation was based on the side and pole side crash test protocols for the KNCAP. With existing safety devices, it can be seen that parts of the occupant's body are not protected by the airbag during the crash. or that the level of injury tends to increase Based on this, we will use this as a reference when developing a new crash evaluation protocol for the KNCAP

Verification is an essential procedure for advancing autonomous vehicle technologies. While validation using real vehicles ensures reliability, it faces limitations in terms of cost and safety. Therefore, simulation-based research utilizing virtual environments has emerged as an effective approach for verifying autonomous driving technologies. However, if virtual vehicle dynamics do not accurately reflect real-world vehicle behavior, validating the reliability and stability of these technologies become challenging. To address this issue, this study proposes a methodology for verifying vehicle dynamics in virtual environments based on actual vehicle data. In the first verification stage, vehicle behavior in the virtual environment is validated using driving input data collected from real vehicles. The second stage involves comparing motion data from actual and virtual vehicles to evaluate their dynamic similarity. Through this two-step approach, vehicle dynamics in virtual environments can be quantitatively assessed, enhancing simulation reliability based on real-world data. The proposed method aims to improve the reliability and efficiency of autonomous vehicle verification, complementing the limitations of physical testing and significantly reducing associated costs and risks.

In accordance with the hydrogen economy revitalization roadmap, significant investments are being directed towards hydrogen electric vehicles, supported by various government policies aimed at resolving environmental pollution issues. The global hydrogen electric vehicle market is experiencing rapid annual growth, driven by substantial government-led investments and deregulation by international competitors. The Korean government is also making efforts to improve diverse regulations within the hydrogen industry to invigorate its leading hydrogen economy and foster a robust hydrogen ecosystem. However, regulations during the development phase remain stringent due to concerns regarding safety and reliability. This study examines the current status and overview of regulatory improvements, along with potential avenues for further enhancement. It proposes optimal improvement directions for pressing regulatory issues that hinder the wider adoption of hydrogen mobility. These issues include allowing modular charging of hydrogen storage systems, harmonizing international standards, and expanding the operation of certification testing and evaluation institutions.