Utilizing Cielab’s virtual data creation solution ‘X-GEN’
Learning autonomous driving situation simulation with virtual data

Seelab's 'X-GEN' will be utilized in a national project to build AI learning data required for accurate object recognition in various indoor and outdoor non-road environments.

Seelab announced on the 23rd that it is constructing 'Delivery Robot Off-Road Operation Data' as part of the '2022 AI Learning Data Construction Project' hosted by the Ministry of Science and ICT and promoted by the National Intelligence Service (NIA).

The goal of this project is to build 500,000 high-quality AI learning data sets (400,000 2D image data sets and 100,000 3D lidar data sets) for commercializing autonomous robots, and 60% of the total data set will be virtual data.

By virtually implementing environments and obstacles where robot driving is impossible through virtual data, the robot can be used for AI learning, and its driving path judgment ability is strengthened.

Seelab's virtual data creation solution 'X-GEN' is used to build virtual data.

X-GEN is software that rapidly generates and augments virtual image data required for AI learning.

Virtual data can be used to simulate and learn about various situations that may occur during robot driving.
/> Seelab has been conducting business since June of this year by forming a consortium with institutions and companies that have acquired expertise and execution experience in each field, such as Naver Labs, Unmanned Solutions, Media Group People and Forest, Bound4, and Korea Robot Industry Advancement Institute, and is currently focusing on the final work, such as data quality verification.

Data built through business will be disclosed through the ‘AI Hub.’

It is expected that this learning data will be used to develop autonomous robot services that operate in indoor and outdoor non-road environments, such as delivery robots, parking, cleaning, and security.