Our laboratory conducts research and development on active safety systems aimed at preventing traffic accidents before they occur. We focus on Advanced Driver Assistance Systems (ADAS) that utilize various sensor information—including cameras, LiDAR, and digital maps—for environment perception.
These systems actively support safe driving through intelligent driving algorithms and vehicle motion control, covering scenarios ranging from normal operation to critical or emergency situations.
Furthermore, we conduct demonstration experiments using both driving simulators and actual experimental vehicle.
In recent years, passenger vehicles equipped with Lane Keeping Assistance Systems (LKAS) have become widely prevalent.
A common LKA method for passenger vehicles involves calculating a target yaw rate based on the concept of a look-ahead point. This method typically ignores the vehicle side slip angle, as it is considered negligible in passenger vehicles.
However, heavy-duty vehicles with long wheelbases generate significant side slip angles; therefore, ignoring these angles during lane keeping control leads to the problem of large tracking errors.
To address this, this study proposes an LKAS control method that considers both steady-state and dynamic (transient) vehicle side slip angles during cornering, and the effectiveness of the proposed method compared to conventional methods has been confirmed.
References:
Pongsathorn Raksincharoensak, Makoto Amemiya, Shunsuke Tsukuda and Yutaka Hamaguchi, Enhancing Automated Lane Keeping Control Performance of Heavy-Duty Trucks Considering Dynamic Characteristics of Vehicle Sideslip Angle, Proceedings of 8th International Symposium on Future Active Safety Technology toward zero traffic accidents (FAST-zero’25), Arles, France, September 23-26, 2025.
While the use of automobiles continues to spread, the increase in transportation-marginalized people (people with limited mobility) has become a significant social issue in Japan. This is driven by factors such as the reduction of public transportation in depopulated areas and the growing number of elderly drivers.
Although automated driving technologies like Lane Keeping Assistance (LKA) and Adaptive Cruise Control (ACC) have been implemented in passenger vehicles, their application to dynamic and narrow environments, such as urban streets, remains a challenge.
In this study, we extended optimization control based on risk potential fields to develop motion control for safely passing oncoming vehicles in narrow roads.
We proposed a method to achieve appropriate passing maneuvers by optimizing the yaw rate and acceleration using Sequential Quadratic Programming (SQP). Furthermore, the effectiveness of the proposed system was confirmed through simulations.
In recent years, significant social issues in Japan include the increasing difficulty of maintaining public transportation in rural areas due to the rapid decline in the birthrate and an aging population, as well as a severe shortage of drivers in the logistics industry.
As a solution to these challenges, expectations for driverless automated driving mobility services are growing. On the other hand, navigating narrow urban streets where pedestrians and bicycles coexist remains technically difficult and poses a major challenge for autonomous driving systems.
This study focuses on the Social Force Model (SFM). The SFM is a classical model used to simulate human movement and crowd dynamics, such as many pedestrians moving toward an exit in scenarios like building evacuations. Based on this model, this research designs an automated driving control system for roads with mixed traffic, including pedestrians and bicycles.
Furthermore, in addition to virtual repulsive forces against traffic participants, we have devised a safer and more natural path planning method that considers the drivable area (boundary constraints). The effectiveness of the control system was verified through experiments using an actual test vehicle.
References:
Shumma Takeda, Yohei Fujinami, Pongsathorn Raksincharoensak, Path Planning to Control Automated Driving Vehicle by Applying Social Force Model, Proceedings of 8th International Symposium on Future Active Safety Technology toward zero traffic accidents (FAST-zero’25), Arles, France, September 23-26, 2025.
This study aims to develop an algorithm to predict the darting-out velocity of pedestrians and other traffic participants based on near-miss incident data collected from real-world driving, and to apply this algorithm to deceleration assistance control.
The system aims to reduce collision risks by decelerating the vehicle to a safe speed by a designated position whenever a blind spot (occlusion) is detected.
To determine appropriate the value of safety speed adapted to various road environments with poor visibility, it is essential to build an algorithm that can calculate the potential darting-out velocity as soon as onboard sensors, such as cameras, recognize the traffic environment context.
This research reproduces near-miss scenarios using a digital twin to verify the effectiveness of the proposed risk-anticipatory deceleration assistance system.
Crossing collisions are characterized by the fact that occlusions (blind spots) often cause delay in the detection of darting-out objects, resulting in a short time margin for avoidance maneuvers. Consequently, existing systems such as Automated Emergency Braking Systems (AEBS) may find it difficult to sufficiently avoid accidents in all cases.
The goal of this study is to realize an intelligent driver assistance system that reduces the risk of crossing collisions by constructing a model to predict the characteristics and dynamic behavior of objects darting out from blind spots.
Based on near-miss incident data of TUAT, we demonstrate the feasibility of estimating the darting-out velocity by probabilistically representing the relationship between the traffic environment context and the behavior of the darting-out object.
References:
Ryuki Ota, Aki Tanikawa, Pongsathorn Raksincharoensak, Masao Nagai, Yohei Fujinami, Takaya Yamazaki, and Minoru Higuchi, “Investigation on Prediction Method of Traffic Participant Darting-out Velocity from Blind Spot Considering Traffic Environment Context,” Proceedings of 2024 JSAE Autumn Congress, Sendai, Japan, October 25, 2024.
Adaptive Cruise Control (ACC), which has become increasingly prevalent in consumer vehicles in recent years, requires continuous driver monitoring. However, there is a tendency for drivers to overtrust the system, which can lead to inattentive driving behaviors or distracted driving. Conversely, it has been reported that the use of ADAS, such as ACC and Lane Keeping Assistance (LKA), can reduce crash risk.
This study aims to elucidate the mechanism behind how ACC influences driver behavior and judgment, and how it contributes to accident avoidance. Through driving simulator experiments, we analyzed driver risk-avoidance behavior in high-risk scenarios while the driver was in a distracted state induced by secondary tasks.
Furthermore, we propose a safer adaptive car-following control system tailored to the driver’s inattentive state and evaluate the acceptability of this driver assistance system.
References:
Norika Arai, Shunosuke Tsujide, Yohei Fujinami, Pongsathorn Raksincharoensak, Fumio Sugaya, Toshinori Okita, Shintaro Inoue, and Masaaki Kamichi, “Analysis of the Effects of Deceleration Maneuvers on Driver Forward Collision Avoidance Behavior during Adaptive Cruise Control Operation,” Proceedings of 2024 JSAE Autumn Congress, Sendai, Japan, October 25, 2024.
N. Arai, S. Tsujiide, Y. Fujinami, P. Raksincharoensak, F. Sugaya, T. Okita, S. Inoue, M. Uechi, Safety Enhancement of Adaptive Cruise Control Adapted to Driver Eyes-Off State, J. Robot. Mechatron., Vol.37 No.5, pp. 1162-1171, 2025.
Norika Arai, Pongsathorn Raksincharoensak, Masao Nagai, Does Automated Lane Keeping System Increase Safety During Distracted Driving?, Proceedings of 8th International Symposium on Future Active Safety Technology toward zero traffic accidents (FAST-zero’25), Arles, France, September 23-26, 2025.
To develop further countermeasures for motorcycle accidents, this study focuses on “right-turn collisions” (accidents involving right-turning cars and straight-going motorcycles), which represent the most frequent accident type. We conduct an analysis using a real-world traffic near-miss incident database.
Furthermore, to understand the driving behavior of car drivers from the collected numerical data in intersection right turn maneuver, we reproduce the corresponding traffic environments on a driving simulator based on near-miss cases extracted from the database.
By analyzing data such as gaze behavior and driving operations through subject experiments, we aim to deepen the understanding of accident causal factors and create defensive driving strategies from a motorcyclist’s perspective.
References:
Tomohito Suzuki, Yohei Fujinami, and Pongsathorn Raksincharoensak, “Research on Countermeasures to Prevent Motorcycle Collisions at Intersections,” Proceedings of the 26th SICE System Integration Division Conference (SI2025), Hiroshima, Japan, December 10–12, 2025, No. 1C1-17.
This study develops a virtual-reality (VR) pedestrian-crossing experimental platform that can safely present high-risk crossing scenarios to support both the education and modeling of decision-making and behavior in older pedestrians. A real-world roadway segment was reconstructed in Unity for this purpose.
Participants physically walked while wearing a head-mounted display (HMD) to decide whether to initiate a crossing, and behavioral data —including walking speed and crossing initiation timing—were collected.
Risk was quantified using time-to-collision (TTC), and statistical analysis revealed that unsafe crossing behavior is more strongly influenced by visual acuity, the perceived realism of the VR environment and approaching vehicles, and walking speed during crossing than by chronological age per se. Furthermore, it was confirmed that improved realism suppresses risky crossings. The VR platform constructed in this study is applicable to crossing education targeting risk perception in the elderly and is effective for the development of individualized crossing-decision models.
This is expected to contribute to the advancement of traffic-safety measures as well as the understanding and evaluation of pedestrian behavior for the design of Advanced Driver Assistance Systems (ADAS) and automated driving systems.
References:
Xingguo Zhang, Hiraki Watanabe, Xun Shen, Pongsathorn Raksincharoensak, Effect of an On-road Crossing Warning System on Pedestrian Safety Using a Virtual Walking Simulator Device, Proceedings of 7th International Symposium on Future Active Safety Technology Towards Zero-Traffic-Accidents (FAST-zero’23), Kanazawa, Japan, 9 November 2023.
In recent years, automotive insurance products linked with drive recorders have emerged, and telematics automobile insurance, which utilizes driving data obtained from drive recorders to prevent accidents, has become widely prevalent.
To realize a safe and secure mobility society without accidents, there is a growing demand to provide services that lead to accident precursor detection and accident prevention by maximizing the use of traffic big data collected via drive recorders, under the new concept of “moving from insurance after an accident to insurance that prevents accidents”.
With the goal of developing next-generation driver assistance systems for accident prevention, this study utilized telematics data provided by the insurance company to analyze the actual conditions of crossing collisions involving vehicles at stop intersections.
Specifically, we analyzed velocity patterns when passing through stop intersections and the appearance positions of crossing vehicles using video image analysis. Furthermore, the accident reduction effect of Automated Emergency Braking Systems (AEBS) was quantitatively estimated.
References:
Pongsathorn Raksincharoensak, Yoko Hojo, Kotaro Seki, Xingguo Zhang, Masami Aga, and Shotaro Yamasaki, “Analysis on Crossing Collision in Stop Intersection by Utilizing Drive Recorder Video Images,” Proceedings of 2025 JSAE Spring Congress, Yokohama, Japan, May 21, 2025, No. 141.
Masami Aga, Yoko Hojo, Pongsathorn Raksincharoensak, Shotaro Yamasaki, and Masao Nagai, “Comparison of Taxi and General Driver Behaviors in Vehicle-to-Vehicle Crossing Collisions at Non-signalized Intersections,” Proceedings of 2025 JSAE Spring Congress, Yokohama, Japan, May 21, 2025, No. 142
COPYRIGHT ©2019 PONGSATHORN LAB. ALL RIGHTS RESERVED.
Page Top