Nvidia presented Nvidia CosmosAn innovative platform designed to accelerate the physical development of artificial intelligence-spat intelligence behind robots, autonomous vehicles (AVS) and other automated systems in the real world. Combining the most modern global Foundation (WFMS) models, advanced video processing tools and data pipeline based on AI, Cosmos allows programmers to create, train and optimize AI models more effectively than ever before.
The development of physical artificial intelligence traditionally required huge amounts of real data, which makes it a costly and temporary process. Nvidia Cosmos aims to change this by offering synthetic generation of data -based data, enabling programmers to create photo report 3D environments that imitate real conditions. These simulated environments help in training AI models without relying entirely on expensive, hand -collected data.
NVIDIA describes the models of the world Foundation as fundamental for the next wave of artificial intelligence, as well as the models of large languages (LLM) revolutionized the processing of natural language. WFM uses a combination of text data, images, video and sensors to simulate interactions in the real world, which makes them necessary for robotics and autonomous systems that must navigate complex environments.
Cosmos includes a number of advanced AI tools adapted to the development of robotics and AVS:
- Generating synthetic data -Using Cosmos, programmers can create high loyalty, video simulations of consciousness of physics of industrial and drive environments, reducing dependence on real data collection.
- Searching and understanding of the video -The ability to search powered by AI allows users to quickly locate specific training scenarios, such as dangerous road conditions or crowded storage environments.
- Predictive intelligence and “MultIVERSUM” simulation -Cosmos can simulate many potential real scenario results, helping AI models to predict the best way of action.
- Advanced data processing – Nvidia curator accelerates the processing of mass video data sets, which makes AI more efficient training.
Cosmos also introduces a visual tokenizer, which can compress and process video data 12 times faster than existing methods, facilitating the conversion of video recordings into useful training data.
Several leading robotic and automotive companies have already begun to integrate Cosmos with AI work flows. Among them are Xpeng, Robotics of agility, AI, Wayve and Uber figure, everyone using the space for the development of new generation AVS and humanoid robots. For example, Waabbi, a company focused on autonomous AI, uses Cosmos for data treatment and AV simulation, while Uber works with NVIDIA to develop autonomous mobility solutions.
Because the content generated by AI becomes more widespread, Nvidia built a cosmos with strong ethical security. The platform includes handrails to prevent the generation of harmful or misleading content, as well as invisible watermarks to identify films generated by AI. The cosmos is compatible with global AI safety initiatives, including the voluntary obligations of the White House.
NVIDIA COSMOS is now available under a model open license to hug the face and the NVIDIA NGC directory. Thanks to the physical artificial intelligence, which is ready to transform industries from production into transport, Nvidia Cosmos means a significant step towards making robotics driven by AI more scalable, efficient and widely available.
Learn more about Cosmos World Foundation model platform for physical artificial intelligenceIn the article available at ARXIV.
