During the #GTC20 Absentee Keynote, NVIDIA CEO and Founder Jensen Huang demonstrated the industry's first robot AI development platform with simulation, navigation, and manipulation.
The presented video shows how the NVIDIA Kaya robot model pushes hard bodies and soft beach balls, the manipulator model manages baskets, trays and other objects, and finally the #robot logistics robot transports a pallet with cargo, taking into account external obstacles and movements of similar robots. Simulations of automated actions made in Isaac SIM 2020.1 are then just as well performed in real conditions.
GPUs accelerate machine learning, perception and planning using deep neural networks. Process modeling allows you to accelerate the development, training and testing of robot algorithms.
The Isaac SDK includes the Isaac Engine application framework, the Isaac GEM high-performance robotics packages, the Isaac Apps reference applications, and the powerful Isaac Sim for Navigation simulation platform. These tools and APIs accelerate robot development by making it easier to implement artificial intelligence for spatial awareness and navigation.
Developer toolset optimized for systems , introduced in December 2018 and providing a good combination of performance and energy efficiency of autonomous machines. JetPack SDK includes NVIDIA CUDA, DeepStream SDK, libraries for deep learning, computer vision, accelerated computing and multimedia.
The Isaac SDK leverages machine learning and continuous testing workflows when running Isaac Sim on NVIDIA DGX systems designed for AI and analytics. These fully integrated solutions are designed to provide data scientists with the most powerful tools for AI and machine learning.
The Isaac SIM 2020.1 Developer Toolkit and Platform will be available later this month.
Source: 3dnews.ru