DistantNews
Support us
🇰🇷 South Korea /Technology

Naver Labs unveils 'DIVINE' AI encoder to make robot brains lighter and faster

From Hankyoreh · () Korean

Translated from Korean, summarized and contextualized by DistantNews.

At a glance

News Named sources New plan
  • Naver Labs Europe has unveiled DIVINE, a new AI encoder designed to make autonomous robots' "brains" lighter and faster.
  • DIVINE integrates various visual AI functions, reducing the need for multiple encoders and significantly cutting memory usage and processing time.
  • The technology promises to lower the barrier for implementing AI in robots for both industrial and everyday applications.

Naver Labs Europe has introduced DIVINE, a versatile artificial intelligence encoder aimed at streamlining the processing capabilities of autonomous robots.

This new technology is designed to act as the "brain" for self-driving robots, enabling them to perceive their surroundings more quickly and accurately while using fewer computational resources. DIVINE consolidates multiple visual AI functions, such as image understanding and spatial and human recognition, into a single, unified encoder. This integration addresses a key inefficiency in current robotic systems where separate encoders were needed for different tasks, leading to redundant data processing, increased memory consumption, and slower performance.

Naver Labs Europe developed DIVINE using a technique called "Multi-Teacher Distillation." This method involves training a single model by distilling the core knowledge from various specialized expert models in areas like image understanding and spatial recognition. The result is a more efficient and powerful AI encoder.

The lightweighting of robot brains is emerging as a major topic for the commercialization of physical AI worldwide.

— Lee Dong-hwan, leader of Naver Labs' Vision GroupDiscussing the significance of DIVINE in the field of AI robotics.

Performance tests have shown significant improvements. Compared to systems using multiple encoders, DIVINE reduced encoder memory usage by approximately 90% and increased encoding processing speed by up to 12 times. The overall robot system's memory usage decreased by about 62%, with processing speed improving up to fourfold. This breakthrough overcomes the limitations of traditional AI models that required high-performance computing environments, paving the way for advanced AI to be implemented in a wider range of robots.

"The lightweighting of robot brains is emerging as a major topic for the commercialization of physical AI worldwide," said Lee Dong-hwan, leader of Naver Labs' Vision Group. "DIVINE will contribute to lowering the barrier to adopting AI robots across daily life and industrial sites."

DIVINE will contribute to lowering the barrier to adopting AI robots across daily life and industrial sites.

— Lee Dong-hwan, leader of Naver Labs' Vision GroupHighlighting the practical applications and impact of the new AI encoder.
DistantNews Editorial

Originally published by Hankyoreh in Korean. Translated, summarized, and contextualized by our editorial team with added local perspective. Read our editorial standards.

More Stories

From Our Blog