Long-range and High-resolution Automotive LiDAR System based on a Stacked SPAD Depth Sensor

Shunpei Suzuki･Takahiro Kado･Ibuki Fujioka･Koji Yamamoto･Shengchao Zheng･Seiya Kaito･Tatsuya Yui･Gyongsok Song･Tomoyuki Taguchi･Taro Beppu (Sony Semiconductor Solutions)

As Advanced Driver Assistance Systems (ADAS) and autonomous driving progress, the demand for LiDAR systems that accurately detect road conditions, vehicles, and pedestrians, even at long distances, is growing. This study presents an automotive LiDAR system that uses a high-sensitivity, stacked SPAD depth sensor. The system achieves long-range detection of up to 300 meters at 10% reflectivity, has a wide field of view (H120° x V25.6°), and high resolution (H0.1° x V0.05°), while allowing the simultaneous acquisition of NIR images with a maximum resolution of 1200×1536 pixels.

Automotive towards SDV - Challenges and opportunities in ADAS&AV features

Xavier Sellart Ortega (IDIADA Automotive)

The automotive industry is rapidly transitioning toward era of SDV's. Bridging the gap between research and current development in ADAS involves translating cutting-edge research findings into practical, scalable solutions. This process requires close collaboration between academia, industry, and regulagory bodies to ensure that emerging technologies are effectively into real world applicaitons. By fostering collaboraiton, sharing knowldege, and investing in robust testing and validation processes, we can accelerate the adoption of advanced ADAS features, enhancing safety and efficiency on the road.

Trustworthy Multimodal Generative AI for ADAS Development

Son Tong･Balakrishnan Ayyanar･Shiye Fang･Kohei Noma･Satoshi Sekine･Theo Geluk･Reiji Takeuchi (Siemens Digital Industries Software)

Generative AI (GenAI) is a powerful technology that learns patterns from data and generates new content, with recent prominent examples of ChatGPT and other large language models (LLMs). This work discusses Multimodal GenAI technologies to accelerate ADAS development involving autonomous driving scenarios and realistic image/video data generation. We use human intelligence, engineering abstraction, and safety requirements as inspiration for trustworthy GenAI models. First, we develop GenAI to reason logged dataset and automatically extracting critical scenarios. Second, methods to generate consistent, controllable, and realistic novel traffic videos will be shown. Finally, we incorporate ADAS standards (EuroNCAP/ISO) to generate simulation-based test cases.

Tractor Driving Simulator for Agricultural Smart Safety

Kenshi Kenshi Sakai Sakai (Tokyo University of Agriculture and Technology)･Masami Matsui･Takahiro Tamura (Utsunomiya University)･Watanabe Masahisa (Tokyo University of Agriculture and Technology)･Keisuke Kazama (Nihon University)･Hidehiko Inoue･Hiroki Takimoto･Ichiro Harada (NARO)･Yuya Aoyagi (Rykyu University)･Pongsathorn Raksincharoensak (Tokyo University of Agriculture and Technology)

Fatal accidents in agriculture are 10 times more frequent than occupational accidents, with tractor rollovers being the leading cause. Strong nonlinearity in the tractor-terrain system causes dangerous behavior. Tractor design, road and field design aim to maximize yield and profit. The tractor driving simulator we developed has been shown to be effective in incorporating "safety" into these design.