Case Study
Monday, September 25
03:30 PM - 04:00 PM
Live in Berlin
Less Details
Two advanced techniques will be explained for driving environmental recognition using millimeter-wave radar. One is the object classification based on domain adaptation technique that leverages simulation data. Accuracy to classify traffic participants such as car, pedestrian and bicycle can be improved with fewer annotated dataset by fusing experimental and simulation data with domain adaptation technique. The algorithm and experimental results using the simulator will be presented. The other one is parking vehicle shape estimation applying semi-supervised learning. The novel semi-supervised learning method is proposed to accurately estimate parking vehicle shapes with fewer ground truths. It is effective because it costs a lot to generate the ground truth. The algorithm and experimental results for the actual parking scenes will be introduced. The audience will learn about:
Tokihiko Akita has been engaged in the research and development of ADAS regarding vehicle control, vehicle dynamics, and surrounding environment recognition for many years in an automotive system supplier. After that, I transferred to the Toyota Technological Institute. I’m now researching autonomous vehicle technology, especially environmental recognition applying AI and millimeter-wave radar. I'm a doctor of engineering and a fellow of the Society of Automotive Engineers of Japan.
The Pop in Your Job – What drives you? Why do you love your topic?
I am a researcher engaging in research and development in this automotive control and AI field for many years. I would like to contribute to the development of this field through my knowledge and various experiences.