NVIDIA has announced its new Cosmos world models, developed for use in robotics and artificial intelligence. The US-based technology company aims to improve the reasoning, visual, and language capabilities of robots and AI agents with these models.

NVIDIA Announces Robotic AI Model

Among the models introduced, Cosmos Reason stands out. An AI model with 7 billion parameters, it enhances robots’ memory and understanding of their physical environment. This allows robots to plan their next moves and reason.

Cosmos Transfer-2, another product in the Cosmos series, accelerates the generation of synthetic data from 3D simulation scenes and spatial control inputs. This simplifies the creation of text, image, and video datasets for training robots and AI systems.

NVIDIA also offers a new visualization technique that allows developers to simulate the real world in 3D using sensor data. This neural system will be integrated into the open-source simulator CARLA to support virtual world creation.

NVIDIA’s new Cosmos models pave the way for significant advancements in robotics and artificial intelligence technologies. These models will allow robots to better understand their environments and perform complex tasks more effectively.

NVIDIA’s Cosmos models will find widespread use in various sectors, particularly autonomous vehicles, industrial robots, and smart devices. This new infrastructure offered to developers will enable robots to adapt to real-world conditions and exhibit more flexible behavior. This will pave the way for the more widespread and effective use of AI-enabled technologies in daily life.