• Session No.51 Safety
  • May 22Room G414+G41512:40-14:20
  • Chair: Daisuke Ito (Kansai University)
For presentations that will not be available video streaming after congress, a “✕” is displayed in the “Video” column, so please check.
No. Video Title・Author (Affiliation)
1

Study on Ventilation Volume of EV in the Case of Transport for COVID-19 Patient (Part4)

Koichi Oshino (non)

Cabin ventilation is important in the case of transport for COVD-19 patients. At the previous report, it is focused on the small opening window case, and it is cleared the relation between opening window height and the ventilation volume by the simple method. Here, this method is applied to the real vehicle, it is examined the relation between opening window height and the ventilation volume.

2

Enhancing Vehicle Safety Through AI-based Motion Prediction and Drivable Space Estimation

Emilia Silvas・Robin Smit (TNO (Netherlands Organisation for Applied Scientific Research))・Manuel Munoz Sanchez (Eindhoven University of Technology (TU/e))・Koichi Kawaguchi (TNO Japan)

This paper presents advanced novel (AI-driven) methods for motion prediction and drivable space estimation, crucial for improving Automated Vehicle (AV) safety, comfort, and efficiency. Results from Dutch and European projects showcase techniques to anticipate crossing pedestrians and aggressive vehicle maneuvers, enhancing situational awareness. These efforts provide insights into AI model requirements and safety. Furthermore, we introduce SDS++, a drivable space estimation framework that adapts to dynamic environments using multi-sensor inputs, surpassing traditional methods. Validated with real-world data and simulations, SDS++ integrates with a model predictive control (MPC)-based planner for context-aware trajectory adaptation, significantly improving AV planning and reliability.

3

A Study on Natural Language-Driven 3D Styling Generation Technology for Vehicle Styling

Akihiko Katagiri・Yoshikazu Nakagawa・Shin Saeki・Jun Shiraishi・Osamu Ito (Honda Motor)

We developed a cutting-edge technology to efficiently design vehicle styling while meeting the requirements of pedestrian protection performance. By using Retrieval-Augmented Generation (RAG) technology, we built a system that enables the surrogate model to be invoked seamlessly through natural language interaction. This system can automatically generate 3D styling shapes. By incorporating this approach, the technology significantly enhances the ability to explore and evaluate numerous styling proposals with high precision during the early stages of vehicle development. This ensures a streamlined process while maintaining both safety standards and attractive design qualities.

4

EU landscape for safety assessment of Connected and Automated Driving

Olaf Op den Camp (TNO)

In the Horizon 2020 and Horizon Europe research and innovation frameworks of the European Commission, large projects have been defined to develop and harmonize methods for the safety assessment and safety assurance of connected and automated mobility systems. The paper will provide an overview of these projects, their results, and their relation with other international initiatives, e.g. in Japan, South-Korea and the USA. This will be put in the context of international regulations and standardization activities.

Back to Top