Development of an artificial weather chamber that reproduces a dynamic weather environment for autonomous driving sensors (4th report)

Hemant SUBHAS Ningaraddi･Tatsuki Yoshida･Hiroyuki Enoki･Hisayasu Shima (ESPEC)･Ifuu Ooishi･Sou Igarashi･Masao Nakagawa (Doshisha University)

To facilitate the extensive adoption of automated driving technologies, novel methodologies for assessing the recognition capabilities of in-vehicle sensors, such as cameras and LiDAR, which are susceptible to the effects of diverse weather conditions, must be developed. This study evaluated the recognition performance of various sensor types, including those employed in automated driving systems, within a fog environment where the size of fog droplets and visibility were controlled using an artificial weather testing chamber.

A Study on SNN-Based Decision Models for Autonomous Driving

Katsuo Semmyo･Katsuki Tomita･Yuya Arima･Shuo Huang (NTT DATA Automobiligence Research Center)

Compared to conventional Artificial Neural Networks (ANNs), Spiking Neural Networks (SNNs), which mimic brain-like structures, offer advantages in low power consumption and low latency, making them promising for applications in automotive functions with demanding hardware requirements. As part of developing a decision model using reinforcement learning, we evaluate an SNN in a simplified experimental environment and discuss its potential.

Physics-Guided Two-Stage Anomaly Detection Framework for Industrial Connector Fitting Verification

Ryusho Itoh･Mako Sasaki･Koichi Ichige (Yokohama National University)･Hirotatsu Matsuo･Atsushi Ugaji･Matsutaro Hata･Kouji Abe (Nissan Motor)

We propose a physics-guided two-stage anomaly detection framework for industrial connector fitting inspection to guarantee zero-defect outflow. Stage 1 feeds CWT-based log-scalograms into a linear bottleneck convolutional autoencoder, suppressing over-generalization and scoring anomalies with a hybrid MSE–SSIM metric. Stage 2 applies a deterministic fail-safe layer evaluating intersection magnitude (IM) with a modified z-score, eliminating remaining false negatives. On five automotive connector types, Stage 1 achieved F1 = 99.36% on the most challenging connector and 100% on all others; Stage 2 completely eliminated false negatives. Under open-set conditions, recall was 99.97% versus 83.37% for CNN baselines.

Correlation between ESA Immunity Test Methods (Second Report)
-Influence of Circuit Impedance and Injection Point-

Shinichiro Itoh･Keiji Kobayashi (NALTEC)･Hitoshi Tsukahara (Japan Quality Assurance Organization)

Comparative measurements were conducted on the correlation between the ALSE method, BCI method, and RVC method, which are ESA immunity test methods specified in ISO 11452. The comparative measurements regarding the effect of different circuit impedances showed that the effect of circuit impedance was relatively small in these three test methods. Furthermore, comparative measurements regarding the effect of different injection points in the BCI method showed that the null point frequency may change depending on the injection point, and that the effect of induced voltage due to the injection point was relatively small.

Investigation of the Impact on Emission Performance during Chassis Dynamometer Testing

Katsumasa Aoki (Toyota Motor)

With changes in the market electromagnetic environment, EMC regulations are becoming stricter and the importance of ensuring EMC performance is increasing. Vehicle EMC emission limits are defined by standards such as CISPR 36, and on-road measurements performed on a chassis dynamometer are required. This paper examines the effect of static electricity generated during chassis dynamometer testing on emission performance.

Development of MILS Technology for Auxiliary Power Supply Control Systems

Takahiro Kinoshita･Yuki Hata (SUBARU)

This paper reports on the overview and achievements of MILS technology development for an auxiliary power supply control system.