NVIDIA Accelerates AI-Enabled Robotics with Simulation and Perception Tools
October 18, 2023
Story
Today, NVIDIA announced a broad series of major updates to its Isaac robotics platform. The announcements came at the ROSCon 2023 event.
NVIDIA said that the platform updates are designed to deliver performant perception and high-fidelity simulation to robotics developers, and they include the release of NVIDIA Isaac ROS 2.0 and NVIDIA Isaac Sim 2023.1, along with upgrades that simplify building and testing AI-based robotic applications for ROS developers.
“ROS continues to grow and evolve to provide open-source software for the whole robotics community,” said Geoff Biggs, CTO of the Open Source Robotics Foundation. “The new NVIDIA prebuilt ROS 2 packages, launched with this release, will accelerate that growth by making ROS 2 readily available to the vast NVIDIA Jetson developer community.”
Isaac ROS 2.0
Isaac ROS 2.0 is reportedly tuned to bring advanced computing to the ROS community at production-quality readiness levels. According to the release, significant enhancements include:
- Native ROS 2 Humble support: NVIDIA will host installation-ready ROS 2 Humble packages for JetPack 5 on Ubuntu 20.04, eliminating the need for source code compilation.
- NITROS ROS bridge: This optimized ROS bridge delivers a 2x speedup compared to standard ROS bridges, extending Isaac ROS acceleration to ROS Noetic applications.
- CUDA NITROS: Empowers developers to use their own DNN models with custom encoders and decoders and harness the acceleration capabilities of NITROS.
- Stereolabs ZED camera integration: Isaac ROS is now integrated with Stereolabs ZED cameras.
- Nova Carter: Isaac ROS software including Nav 2 navigation stack now supported on Nova Carter, a reference AMR for robotics R&D, powered by NVIDIA Jetson AGX Orin.
- ESS 3.0 performance: The new ESS 3.0 depth perception DNN model offers improved accuracy and performance.
Isaac Sim 2023.1
Built on NVIDIA Omniverse, Isaac Sim is a robotics simulator for developing, training, testing, and deploying AI-enabled robots, NVIDIA said. Its capabilities include NVIDIA Omniverse Replicator for generating synthetic data, and Isaac Gym for GPU-accelerated reinforcement learning. In addition, the announcement indicates that the latest release of Omniverse Replicator 1.10 is designed to help developers boost synthetic data generation with a low-code, YAML-based configurator and asynchronous rendering. Isaac Sim reportedly also includes accurate sensor simulations for most of the popular sensors available today. It supports ROS and ROS 2 and can additionally be controlled from a Python script.
Ken Briodagh is a writer and editor with two decades of experience under his belt. He is in love with technology and if he had his druthers, he would beta test everything from shoe phones to flying cars. In previous lives, he’s been a short order cook, telemarketer, medical supply technician, mover of the bodies at a funeral home, pirate, poet, partial alliterist, parent, partner and pretender to various thrones. Most of his exploits are either exaggerated or blatantly false.
