TURING Self-driving EV developer Turing develops traffic light recognition algorithm using AI technology

Self-driving EV developer Turing develops traffic light recognition algorithm using AI technology
Compatible with traffic signals in Japan by using a database developed in-house
TURING Co., Ltd. (Kashiwa City, Chiba Prefecture, CEO: Kazunari Yamamoto, hereinafter referred to as “Turing”), which is engaged in the development of fully automated driving and EV vehicles using artificial intelligence, has developed a high-precision identification of traffic lights using a driving database developed in-house. We have started to develop the model. Turing has completed the acquisition of approximately 1,000 hours of driving data and the construction of a driving database. Using this database and in-house developed AI technology, we are developing traffic signal recognition technology through image recognition. increase. By incorporating traffic light recognition technology based on AI and deep learning into an autonomous driving system, we aim to realize a traffic light recognition system that does not require additional sensor equipment or road infrastructure equipment.
■ Background on the development of traffic signal recognition technology Aiming for fully automated driving, Turing has been conducting research and development activities aimed at realizing an AI automated driving system that more closely resembles human judgment and brains. In order to realize an autonomous driving system, recognition and understanding of the meaning of traffic lights were very important factors. (3) The understanding of the meaning of traffic signals is complicated in the first place. It has not been developed.
We also considered installing additional sensor equipment and road infrastructure equipment for recognition and identification of traffic lights, but in addition to suppressing the introduction cost, it can be used in any place and in any situation regardless of the amount of infrastructure investment in each region. In order to realize possible autonomous driving, we decided to develop a technology that recognizes traffic lights only with a camera attached to the vehicle.
[Image 1

■ About Turing’s development of traffic signal identification technology Turing Corporation is developing a traffic light recognition algorithm to realize traffic light recognition by AI in Japan. For traffic signal identification using AI technology and camera images, we annotated and labeled 15,000 images and developed a traffic signal identification model. For a more detailed development process, the development team has posted a technical article, so please take a look.
[Automatic driving] Challenging traffic light recognition / Deep learning practice with annotation of 15,000 driving images and YOLOX model
By developing and implementing a signal recognition algorithm using AI technology and cameras, we aim to realize a low-cost, scalable automated driving system that does not require additional investment in transportation infrastructure.
About the “Brain Team” that develops Turing’s AI self-driving technology Including Director of AI Yu , engineers with expertise in machine learning, MLOps, cloud, image recognition, etc. are working on development and research for the realization of fully automated driving. The Brain team is working on a research and development project and a driving data collection project to realize a machine learning-based automated driving system. We will continue to actively carry out development activities to realize fully automated driving. [Image 2d98132-15-c3aa7e89c5b1ac5b9947-1.jpg&s3=98132-15-d97ca2e637fd3f733a713b925d0380d6-1999x1334.jpg
■ About Turing
TURING is a startup aiming for mass production of fully autonomous driving EV with the mission of “We Overtake Tesla”. Co-founded in 2021 by Kazunari Yamamoto, the developer of the world’s first shogi AI “Ponanza” that defeated a master, and Shunsuke Aoki, who studied autonomous driving at Carnegie Mellon University and obtained a Ph.D. We aim to realize “fully automated driving” that goes beyond limited areas using AI deep learning technology.
In addition, we have already raised 1 billion yen in seed rounds, but we plan to raise Series A funds in 2023 in anticipation of building a vehicle production system in-house.
Company name: TURING Inc. (Reading: Turing, English notation: TURING Inc.) Representative: Kazunari Yamamoto, Representative Director
Established: August 2021
Capital: 30 million yen (as of the end of September 2022)
Business: Development and manufacturing of fully autonomous EVs Headquarters: 226-44 Wakashiba, Kashiwa City, Chiba Prefecture Central 141 Block 1
URL: https://www.turing-motors.com
■ Recruitment information
We are actively recruiting colleagues to create fully automated driving systems and EV vehicles together.
Recruitment page: https://herp.careers/v1/turing
■ Inquiries from the media
Public Relations (Tanaka/Ono): pr@turing-motors.com

Details about this release:



%d bloggers like this: