Listen to this article
Today, NVIDIA announced a number of new tools for robotics developers during the GTC event, including Jetson Orin Nano system-on-modules, updates to the Nova Orin Autonomous Mobile Robotics (AMR) reference platform and the cloud-based availability of its Isaac Sim technology.
NVIDIA has expanded its Jetson lineup with the launch of the Jetson Orin Nano system-on-modules for entry-level AI and robotics applications. The new Orin Nano delivers up to 40 trillion operations per second (TOPS) of AI performance, which NVIDIA said is 80 times the performance over the previous generation, in Jetson’s smallest form factor to date.
The Jetson Orin features NVIDIA Ampere GPU architecture, ARM-based CPUs, next-generation deep learning, vision accelerators, high-speed interfaces, fast memory bandwidth, and multimedia sensor support. Orin Nano units will be available in two versions. The Orin Nano 8GB offers up to 40 TOPS with configurable power from 7W to 15W, while the 4GB version offers up to 20 TOPS with power options as low as 5W to 10W.
Orin Nano is powered by the NVIDIA JetPack Software Development Kit and is powered by the same NVIDIA CUDA-X Accelerated Computing Suite used to create AI products in areas such as industrial Internet of Things, manufacturing, smart cities, and more.
Jetson Orin Nano units will be available in January 2023 starting at $199.
“More than 1,000 customers and 150 partners have adopted the Jetson AGX Orin since NVIDIA announced its availability just six months ago, and Orin Nano will greatly expand that adoption,” said Deepu Tala, Vice President of Embedded and Advanced Computing at NVIDIA. “With a significant increase in performance for the millions of advanced AI and ROS developers today, Jetson Orin is the ideal platform for nearly every type of bot deployment imaginable.”
Orin Nano modules are form factor and compatible with previously announced Orin NX modules. Full simulation support allows customers to begin development for the Orin Nano series today using the AGX Orin developer suite. This gives customers the flexibility to design a single system to support multiple Jetson modules and easily scale their applications.
The Jetson Orin platform is designed to solve tough bot challenges and deliver accelerated computing to more than 700,000 ROS developers. Combined with the powerful hardware capabilities of the Orin Nano, improvements in the latest NVIDIA Isaac software for ROS put increased performance and productivity into the hands of the bots.
Jetson Orin has seen extensive support across bots and the embedded computing ecosystem, including from Canon, John Deere, Microsoft Azure, Teradyne, TK Elevator, and many more.
Isaac Sim in tow
NVIDIA also announced that there will be three ways to access it Isaac Sim robotics simulation platform on the cloud:
- Coming soon on the new NVIDIA Omniverse Cloud Platform
- It’s now available on AWS RoboMaker
- Developers can now download it from NVIDIA NGC and publish it to any public cloud
Roboticists will be able to create large datasets from physically simulating micro-sensors to train AI-based cognition models on their robots. The synthetic data generated in this simulation improves model performance and provides training data that often cannot be collected in the real world.
NVIDIA said the upcoming release of Isaac Sim will include NVIDIA cuOpt, a real-time fleet task assignment engine and route planning engine to improve robot route planning. By leveraging the accelerated performance of the cloud, teams can make dynamic, data-driven decisions, whether it’s designing the perfect repository or optimizing active operations.
Nova Oren Updates
NVIDIA has also shared new details about the three Nova Orin reference platform configurations for AMR devices. Two use the Jetson AGX Orin one—which runs the NVIDIA Isaac botnet stack and the Robot OS (ROS) with a GPU acceleration framework—and the other is based on the two Orin units.
NVIDIA said the Nova Orin platform is designed to improve reliability and reduce development costs for building and deploying AMR. Nova Orin offers industry-grade configurations of sensors, software, and GPU computing capabilities.
Nova Orin reference architecture designs are provided for specific use cases. There is one Orin-based design without certified safety sensors, and one that includes them, along with a programmable logic controller for safety. The third architecture has a dual Orin-based design that relies on AI vision to enable functional safety.
Sensor support for stereo cameras, LiDAR, ultrasonic sensors, and inertial measurement units is included. The sensors selected were designed to balance performance, price, and reliability for industrial applications. Stereo cameras and fisheye cameras are custom built by NVIDIA in coordination with camera partners. All sensors are calibrated and synchronized with time, and come with drivers to reliably capture data. These sensors allow AMR devices to detect objects and obstacles across a wide range of situations while enabling Simultaneous Localization and Mapping (SLAM).
NVIDIA offers two LiDAR options, one for applications that do not need functional safety-certified sensors, and one for those that require it. In addition to 2D LiDARs, Nova Orin 3D LiDAR supports mapping and collection of ground truth data.
The core operating system includes drivers and firmware for all hardware and adaptive tools, as well as design guides for integrating them with bots. Nova can be integrated with the ROS-based bot application. The sensors will have models that have been validated in Isaac Sim to develop and test applications without the need for a real robot.
Cloud-native data acquisition tools eliminate the tedious task of preparing data pipelines for the massive amount of sensor data needed for training models, debugging, and analytics. State-of-the-art GPUs developed for Nova sensors are accelerated using the Jetson Orin platform, providing building blocks such as visual SLAM, stereoscopic depth estimation, obstacle detection, 3D reconstruction, semantic segmentation, and position estimation.
Nova Orin supports secure updates over the air, as well as device management and monitoring to enable easy deployment and reduce maintenance cost. Its open modular design allows developers to use and extend some or all of the platform’s capabilities to rapidly develop bot applications.
NVIDIA works closely with regulators to develop vision-enabled security technology to reduce cost and improve the reliability of AMRs.
#NVIDIA #launches #bot #development #tools #GTC