Go to Contents Go to Navigation

(LEAD) Hyundai Mobis to utilize 3-D game technology to develop autonomous driving system

All Headlines 17:04 September 16, 2018

(ATTN: UPDATES with more details in last 3 paras)

SEOUL, Sept. 16 (Yonhap) -- South Korea's largest auto parts maker, Hyundai Mobis, said Sunday that it plans to make full use of 3-D mobile game software in the development of its autonomous driving system.

One of the flagship companies of Hyundai Motor Group said using real, high-definition video game images to create all kinds of driving conditions can allow it to more efficiently and safely check the cameras that are an integral part in any self-driving car system.

The videos can be designed to simulate night and bad weather conditions, as well as complicated urban driving environments with lots of pedestrians and other cars, the company said. It can, moreover, be used to check if the cameras can detect and process commonplace repairs being carried out on roads.

"By getting the camera to detect and read all manner of signs and road-related movements and developments, advances can be made to the overall system at a faster pace than through more conventional means," it said. "This arrangement can give Mobis an edge in developments."

If such simulators are not used, a car fitted with cameras and sensors will have to be driven on real streets, taking more time and posing safety issues.

Mobis said that its research center in India has teamed up with Tata Elxsi -- an artificial intelligence, Internet of Things and big data service provider based on the subcontinent -- to develop the necessary 3-D images and software.

The contract calls for the simulation program to be developed by late 2019, Hyundai Mobis said.

Besides building a simulator for the camera system, Hyundai Mobis said it it is in the process of developing an automatic video recognition and categorization system using deep learning technology.

Once perfected, this could greatly enhance the ability of the camera to recognize the images it captures, such as traffic lanes, people, signal lights and various road signs, and make sense of them.

"At present, classifying images requires countless manhours, yet the workload and the time it takes to classify images can be greatly reduced through a deep learning process," a company source said.

This Hyundai Mobis provided image shows a 3-D game image that can be used in the development of self-driving car programs. (Yonhap)


Issue Keywords
Most Liked
Most Saved
Most Viewed More
Send Feedback
How can we improve?
Thanks for your feedback!