Ensuring an automated fallback strategy in response to malfunctions during the execution of the Dynamic Driving Task (DDT) is imperative for Level 4 autonomous driving systems. While Triple Modular Redundancy (TMR) represents a prominent Fail-Operational structure, its practical application to multiple systems is constrained by the substantial increase in costs. In this paper, we propose a pragmatic Fail-Operational safety concept utilizing on-board camera sensors and the Vehicle-to-Infrastructure (V2I) communication module, known as the On-Board Unit (OBU), to provide traffic signal information within the vehicle. The viability of the designed safety concept is validated through error injection simulations. This approach addresses the practical limitations associated with applying Fail-Operational functionality to numerous systems due to the considerable cost escalation. Leveraging camera sensors and V2I communication modules presents a practical and cost-effective solution for maintaining operational safety in Level 4 autonomous driving systems, particularly when responding to malfunctions in the DDT.

Recent advancements in vehicle driving data analysis have attracted significant attention as a key research area aimed at optimizing driving behavior and enhancing vehicle performance. This study aims to collect operational data from drivers, including steering angle, longitudinal acceleration, brake usage, and wheel speed, and to cluster driving segments based on their characteristics using deep learning techniques. To achieve this, we employed a methodology that integrates the LSTM-Autoencoder model, which captures essential features of time-series data through compression and reconstruction, with an Attention mechanism. The driving segments were clustered and visually represented with distinct colors for clarity. Experimental results from the proposed approach demonstrated an over 96% accuracy in aligning input and output values, facilitating the clustering of driving segments based on their distinctive features. These findings provide valuable insights for improving driver behavior and optimizing vehicle assistance systems, potentially contributing to significant advancements in this field of research.

In this paper, we propose an improved event data recorder (EDR) storage data that can determine the cause of unintentional acceleration of a vehicle. The proposed EDR storage data includes whether the mechanical parking brake is applied or not and whether the electronic parking brake (EPB) is applied or not. To validate the proposed EDR storage data, the types, configurations, and main functions of the EPB were studied to verify that the driver can stop the vehicle through dynamic braking by the EPB while driving, and the vehicle speed, the wheel speed of each wheel, and the vehicle deceleration data were obtained through CAN while conducting a practical test under low mue conditions.

Braking judder vibration caused by thermal deformation of disc has been a major problem in brake system for a long time and many researchers have analyzed its mechanisms and developed solutions. However, judder vibration still occurs due to harsher vehicle driving conditions like increased power of EV (Electric Vehicle) and various environmental characteristics. In particular, in the case of eco-friendly vehicles such as EV, it is predicted that judder vibration will become a bigger problem due to the quiet driving condition compared to ICE (Internal Combustion Engine) vehicles. In addition, existing studies on judder vibration have been focused on the capacity and thermal deformation of the braking friction surface. So, the influence analysis of thermal deformation on the non-friction surface of the brake disc is relatively insufficient. In this study, we attempt to secure braking characteristics that are insensitive to thermal deformation in terms of the non-friction surface of the disc, focusing on the coning characteristic that occurs during braking thermal deformation. For this purpose, various factors of the non-friction surface of the disc are analyzed using robust design. The design standard for the robustness of the brake disc against judder vibration is proposed through the research results.

Due to its good repeatability and reproducibility, sled test is an effective way to study the occupant behavior during car crash but sled test is not suitable to simulate long term events (>1000 msec) such as autonomous emergency braking (AEB) which accompanies over 1 second of braking stage before crash. The behavior of dummies at different AEB levels are assessed using the Hybrid-III 50% male dummy and 5% female dummy FE models. Behavior of dummies are studied at different postures like normal posture, relaxed posture, and long-sliding relaxed posture. Dummy behavior from prepositioning (presetting the dummy as dummy will be positioned after AEB activation) and that from full event simulation which includes AEB activation as well as crash will be compared to check the effectiveness of prepositioning. Furthermore, based on the observation that the forward excursion of dummy head remains at a certain level with the lapse of time, we evaluated the applicability of time-shortened pulses method as well. Prepostioning method can be applied to normal posture, but it is not applicable to the relaxed posture, and it is confirmed that it can be corrected by applying the time-shortened pulses method.

MDB (Moving Deformable Barrier) is widely used over the world for the tests like frontal offset impact, side impact as well as frontal oblique impact. As autonomous vehicles are expected to come into the market in near future, new test protocols are required to secure the safety of the new type vehicles. MDB can be a versatile option for the new tests. In order to better understand the MDB, NHTSA oblique impact is chosen for the case study. NHTSA oblique impact test is about to be published and will be one of the newest test protocol using MDB, which bears technical know-how's built up from previous studies of MDB impacts. Compressive deformation behavior of the honeycomb of MDB is studied changing thickness of honeycomb structure. Delta-V is a good measure of impulse to the vehicle and is compared with varying weight of the MDB as well as its initial impact speed. Equivalence of delta-V with respect to momentum and kinetic energy is studied as well.

As autonomous driving performance, such as automatic emergency braking (AEB) and advanced driver assistance systems (ADAS), continues to improve, collision angles and occupant seating postures become more diverse, and there is a need to study how occupant injury mechanisms change depending on the type of collision. Accordingly, a representative crash test mode was derived. Using the derived crash test mode, we analyzed the crash injury mechanism according to the impact angle and the occupant's seating posture (seat back angle). Sled is a crash simulation test that applies a pulse corresponding to the vehicle body acceleration pulse generated during a collision. Sled testing has advantages in terms of cost and time compared to actual vehicle crash testing. We focus on the correlation between crash tests reflecting autonomous vehicle crash modes and Sled tests. The results obtained through this study can be used to develop new crash evaluation methods. As a result, we will present the results of an experimental study on the actual vehicle crash test Sled test method.

With the advent of Level 4 autonomous vehicles, the need for effective communication between these vehicles and pedestrians has become increasingly important. To address this, eHMI (Enhanced Human-Machine Interface) technology has been proposed to replace traditional driver-pedestrian interactions. eHMI plays a crucial role in conveying the vehicle's status and intentions to pedestrians, thereby improving interaction. Globally, automobile manufacturers and technology companies are investing in visual eHMI technologies, advancing in tandem with the automotive industry. This study developed a methodology for field evaluation of communication technologies between pedestrians and Level 4 autonomous vehicles in urban settings. A three-stage message and display system, tailored to the pedestrian crossing process (recognition, judgment, response), was established. In experiments without message displays, 42.2% (38 out of 90) of participants abandoned crossing. Most who crossed did so only after the vehicle stopped, with some groups crossing irrespective of vehicle approach. When the 'yield' message was introduced, crossing patterns and speed distributions changed significantly. All 38 participants who initially abandoned crossing decided to cross, and the elderly who previously ran or walked quickly crossed at a normal pace, reducing overall crossing time. Field experiments are crucial as real-world conditions may elicit different behaviors than controlled settings. Continued research in field evaluations is essential to develop and assess effective eHMI messages. By observing and analyzing actual pedestrian movements, we can reliably evaluate the effectiveness of these communication technologies, thereby enhancing interaction between autonomous vehicles and pedestrians.

As the importance of eco-friendly carbon neutrality is emphasized around the world, hydrogen-electric vehicles are expected to grow rapidly, and as of July 2003, Korea has 32,500 hydrogen cars and 363 hydrogen vans. Hydrogen cars have the advantage of short charging time and long mileage. However, the safety issue of internal pressure-resistant tank is still one of the important challenges in this field. To improve this, the establishment of a safety value chain to strengthen safety is the key. In this study, a method of developing and applying a safety value chain was studied to reinforce the safety of hydrogen-resistant tank. This study is expected to contribute to the expansion of safe and reliable hydrogen supply by actively applying these safety strategies to the hydrogen electric vehicle industry.

The study aims to introduce three new lighting device technologies designed to improve the visibility and awareness of stopped school buses, especially during nighttime, to enhance the safety of children boarding and disembarking. Recent accidents involving various school transportation vehicles underscore the need for effective child safety signals. Analysis of traffic accident data from the Korea Road Traffic Authority's system indicates significant differences between daytime and nighttime incidents. Existing research suggests current lighting devices are inadequate in safeguarding children. Therefore, the study seeks to propose new technologies that better alert surrounding drivers to stopped school buses, compared to existing devices, and explore factors influencing the adoption of technologies like road projections, VMS, and line lamps.

This study is to review the Brain Injury Criterion (BrIC) and the brain injury probabilities due to the Traumatic Brain Injury (TBI). For this purpose, the frontal impact test results of 20 vehicles from MY2022 and MY2023 USNCAP are utilized for estimating the BrIC and the brain injury probabilities. Using the equations suggested by Takhounts et al., the BrIC results were calculated. Using the equations suggested by Hazay et al., the brain injury probabilities were estimated. The brain injury probabilities based on the maximum principal strain (MPS) were largely estimated than those based on the cumulative strain damage measure (CSDM).

In this study, it was proposed to apply the existing driver liability and compulsory automobile insurance system applied to general cars and Level 3 automated vehicles in relation to the responsibility of autonomous driving, ensuring safety, and strengthening accident liability, and to establish a definition of an autonomous driving supervisor who remotely supports autonomous driving and to purchase liability insurance. In addition, in the case of an autonomous vehicle accident caused by the intentional or negligent act of the autonomous driving supervisor, an amendment was proposed to allow insurance companies to compensate for damages to the person legally liable for damages, and an amendment to define the autonomous driving supervision insured and add it as an insured.