From NVIDIA: 2024-10-22 03:01:00

At ROSCon, NVIDIA and partners introduced generative AI tools, simulation, and perception workflows for ROS developers. New AI nodes and workflows were revealed for the NVIDIA Jetson platform, enabling robots to perceive, communicate, and make adaptive decisions autonomously. ReMEmbR, powered by generative AI, enhances robotic reasoning and action using large language and vision models.

ROS developers can leverage WhisperTRT ROS 2 node for speech recognition, utilizing NVIDIA TensorRT to optimize OpenAI’s Whisper model for low-latency inference on Jetson. Projects like ROS 2 robots with voice control and NASA’s ROSA AI agent demonstrate advancements in human-robot interaction and AI capabilities in NVIDIA Isaac Sim.

Canonical showcases NanoOWL, enabling zero-shot object detection on Jetson Orin Nano, while Isaac ROS 3.2 enhancements focus on robot perception, manipulation, and environment mapping. NVIDIA Isaac Manipulator and Perceptor offer improved workflows for pick-and-place operations and environmental awareness in autonomous mobile robots.

Innovative collaborations see robotics companies adopting NVIDIA Isaac for AI-powered applications. Universal Robots launches an AI Accelerator toolkit, Miso Robotics uses Isaac ROS for automation, and Wheel.me partners with NVIDIA for a production-ready AMR solution. Orbbec, LIPS Corporation, and Main Street Autonomy deploy Isaac Perceptor for sensor calibration and AMR navigation.

ROSCon brings ROS community members and partners like Canonical, Ekumen, and Siemens together for workshops and sessions in Odense, Denmark. Highlights include discussions on AI integration, large-scale factory automation, and accelerated robot learning in simulation. NVIDIA and Teradyne Robotics co-host a networking event, showcasing the industry’s latest advancements in robotics and AI integration.



Read more at NVIDIA: NVIDIA Unveils Generative AI Tools, New Simulation and Perception Workflows to ROS Developer Ecosystem