Published on October 18, 2023, 11:17 am
Nvidia, a leading technology company, has recently made significant advancements to its Jetson platform for edge AI and robotics. One of the major developments is the introduction of generative AI for industrial robots, manufacturing systems, and other edge devices.
Generative AI on robots and edge devices allows for learning with minimal or no examples. For instance, vision transformers can now easily identify new objects without the need to train a new model. This breakthrough enables systems to become more adaptive and responsive.
In addition to this, Nvidia’s Jetson platform has gained over 10,000 customers including prominent companies like Amazon Web Services, Cisco, PepsiCo, and Siemens. The platform continues to evolve and cater to the growing needs of the AI industry.
To further support in-house developers, Nvidia has launched the Jetson Generative AI Lab. This platform provides developers with access to the latest open-source generative AI models specifically optimized for the Jetson platform.
Furthermore, Nvidia has updated its TAO toolkit for edge model optimization with VisualChangeNet – a new transformer model designed for defect detection. The company also highlighted that its hardware system, Nvidia Orin, is capable of running foundation models such as Meta’s LLaMA 2 with up to 70 billion parameters.
Nvidia aims to simplify the development of AI-based robotics applications through enhancements in perception and simulation with their Isaac ROS 2.0 and Isaac Sim 2023.1 robotics and simulation frameworks. They are now generally available and offer improved capabilities in these areas.
As part of their commitment to advancing vision AI applications in industries, Nvidia plans to enhance their industrial platform called Metropolis by introducing new APIs and microservices by year-end.
Lastly, Nvidia intends to release version 6.0 of their JetPack software development kit (SDK) called JetPack 6.0. This update will bring numerous new features for developers working on autonomous machines. Deepu Talla, the Vice President and General Manager of Autonomous Machines at Nvidia, mentioned that these updates are a result of years of dedicated work. To assist customers in simplifying their development processes, Nvidia is also providing complete reference workflows for various applications including video surveillance, robotic inspection, and mobile robots.
Nvidia’s continued dedication to advancing generative AI and their comprehensive suite of tools and platforms demonstrate their commitment to driving innovation in the field of edge AI and robotics.