Naver has made a significant announcement that could solve a core problem in robotics technology.
At the heart of this news is DIVINE, a unified 'robot brain' developed by Naver Labs Europe. In simple terms, robots often need separate AI models, or encoders, to understand different things like recognizing objects, estimating distances, and navigating spaces. DIVINE consolidates all these separate functions into a single, highly efficient model. The results are impressive: encoder memory usage dropped by 90%, and processing speed increased by up to 12 times in internal tests.
So, why is this so important? It directly tackles a major hurdle in the field of 'Physical AI'—AI that interacts with the real world. Robots, especially smaller service robots, run on limited battery power and have tight memory constraints. Running multiple complex AI models simultaneously is a huge drain on these resources. This is where DIVINE comes in. By unifying the encoders, it frees up critical memory and computing power, allowing the robot to think and act faster on the device itself, a concept known as edge computing.
This development didn't happen in a vacuum. First, recent academic research highlighted performance bottlenecks on popular robotics hardware like NVIDIA's Jetson platform, underscoring the need for more efficient software. Second, hardware supply chain issues have made memory components more expensive, making a memory-saving solution like DIVINE even more valuable. Finally, the Korean government's supportive policies, like the updated Intelligent Robot Act, are lowering the barriers for deploying service robots in public spaces.
From a business perspective, DIVINE is a game-changer for Naver. The company already operates over 100 'Rookie' robots at its headquarters using a cloud-robot system called ARC. DIVINE can be rolled out to this entire fleet through a simple software update, instantly improving their performance without costly hardware changes. This could significantly lower the manufacturing cost, or BOM (Bill of Materials), for future robots. Despite this promising technology, Naver's stock price actually fell on the day of the announcement. This was likely due to broader market instability, as the Korean stock market (KOSPI) was experiencing significant volatility at the time. This suggests the market has not yet priced in the long-term potential of Naver's robotics division.
- Physical AI: Artificial intelligence that enables a machine, like a robot, to perceive, reason, and interact with the physical world.
- Encoder: A component of an AI model that processes raw input data (like an image from a camera) and converts it into a mathematical representation that the rest of the model can understand.
- BOM (Bill of Materials): A comprehensive list of the raw materials, components, and assemblies required to manufacture a product.