Nvidia unveiled new software that makes it easier for humanoid robots to navigate their environment. Huang, the company's CEO, said that comprehensive use of robots is less than five years away.
At the GTC 2025 event, tech giant Nvidia unveiled Groot N1, a groundbreaking AI base model designed specifically for humanoid robotics. This new model aims to increase the versatility of AI in industrial and personal robotics uses.
Groot N1 is characterized by an innovative “fast and slow thinking” system architecture. In this architecture, the fast thinking system gives robots the ability to react reflexively and instantly. On the other hand, the slow thinking system enables robots to analyze the environment and make more strategic decisions in more complex tasks. Thus, robots can both move quickly and develop a more detailed understanding of the situation.
FAST THINKING SYSTEM
This system enables robots to react reflexively and instantaneously. Robots with it can quickly sense an obstacle that suddenly appears while walking and change direction. They can instantly adjust their grip strength to prevent an object from slipping while gripping it, or take reflexive actions such as quickly retreating when a liquid is spilled in their direction.
The new system allows the robot to quickly adapt to environmental changes and generally operates with low energy consumption.
SLOW THINKING SYSTEM
This system was developed to enable robots to make more complex, methodical decisions. This gives humanoid robots features such as task planning and environmental perception. For example, when you instruct a robot with Groot N1's slow-thinking system to tidy up your room, it first scans the perimeter of the room and makes a detailed mapping, then analyzes the most efficient way to organize objects and performs the task.
Slow thinking enables robots to perform tasks that involve multiple steps. It requires more processing power and energy, but allows the robot to act more deliberately and strategically.
EXAMPLES OF WORKING TOGETHER
These two systems often work in an integrated way. A robot can grasp an object with fast thinking and then decide where to place it with slow thinking. On a production line, fast thinking can detect and sort out defective products, while slow thinking can analyze general problems in the production process.
The fast and slow thinking systems were inspired by Nobel Laureate Daniel Kahneman's “System 1 and System 2” theories on the human mind. The Groot N1 artificial intelligence model adapts nu theory to robotics, enabling humanoid robots to work more naturally and effectively.
INVESTING IN OPEN SOURCE
Jensen Huang, CEO of Nvidia, described Groot N1 as “the beginning of the era of generalist robotics”. The company is not only offering Groot N1, but also releasing open-source synthetic datasets and simulation frameworks where researchers and developers can build their own applications. This creates huge potential for collaboration in the industry.
WIDE RANGE OF USES
The Groot N1 promises a wide range of applications for humanoid robots, from home automation and healthcare to industrial manufacturing and logistics. Nvidia offers flexibility to ensure that the model is compatible with different types of humanoid robots, not just tailored to a specific robot.
“THE ROBOT REVOLUTION IS IMMINENT”
Huang said that he thinks there are less than five years left before robots are used extensively in manufacturing plants. Stating that robots should be in factories first, Huang said that the safety rules of the factory environment are clear and their intended use is clearly defined.
Huang added that the estimated cost of renting a humanoid robot could be 100 thousand dollars.
0 Comments: