Jan 13: Jetson AI Lab Research Group Call - Accelerating Robotics with Isaac ROS on Jetson

| Podcasts | May 04, 2026 | 111 views | 54:17

TL;DR

NVIDIA's Isaac ROS team explains how their NITROS framework eliminates costly GPU memory copies in ROS 2 to enable a new era of "Physical AI" where end-to-end learned policies replace traditional robotic control, requiring tight integration of accelerated computing from simulation to deployment on Jetson.

Isaac ROS & The NITROS Breakthrough 3 insights

The GPU memory copy bottleneck

Standard ROS 2 messages require CPU-only memory, forcing expensive GPU-to-CPU copies between accelerated nodes that create performance 'speed bumps' and GPU idle time.

Zero-copy transport with NITROS

NVIDIA Isaac Transport for ROS uses type adaptation to pass GPU memory handles between nodes, avoiding copies while maintaining compatibility through automatic conversion to standard messages when necessary.

Democratizing accelerated development

PyNITROS and CUDA with NITROS extend zero-copy capabilities to Python nodes and custom CUDA development, allowing developers to build GPU-accelerated components without memory transfer penalties.

🤖 The Physical AI Revolution 3 insights

Three eras of robotics evolution

The industry is transitioning from deterministic automation and intelligent analytic robots to general-purpose robots using end-to-end 'pixels to torques' learned policies.

Humanoids as the ultimate challenge

Humanoid robots represent the epitome of the third era because their complexity makes analytic control impractical, requiring transformer models trained in high-fidelity simulation.

Three-computer workflow

Physical AI development requires robots to be 'born in simulation' using Omniverse, trained on DGX systems, and deployed to Jetson AGX for real-time end-to-end control loops.

🔮 Future of Accelerated Robotics 2 insights

Making ROS 2 accelerator-native

NVIDIA is collaborating with OSRF to integrate NITROS into ROS 2 Lyrical via RCL Buffer, aiming to make ROS 2 'the PyTorch of robotics' by being accelerator-aware yet hardware-agnostic.

Real-time CUDA control loops

Future development targets integrating CUDA into high-frequency (500Hz-2kHz) real-time control loops, referred to as 'SICKLE', necessary for end-to-end autonomy stacks.

Bottom Line

Adopt Isaac ROS with NITROS now to eliminate GPU memory copy overhead and prepare your robotics stack for the Physical AI era where end-to-end learned policies replace traditional analytic control.

More from NVIDIA AI Podcast

View all
Apr 14 - Jetson AI Lab Research Group Call - Tensor RT Edge LLM on Jetson & Culture
51:38
NVIDIA AI Podcast NVIDIA AI Podcast

Apr 14 - Jetson AI Lab Research Group Call - Tensor RT Edge LLM on Jetson & Culture

NVIDIA researchers Lynn Chai and Luc introduce TensorRT Edge LLM, a purpose-built inference engine for deploying large language models on Jetson edge devices, showcasing NVFP4 quantization and speculative decoding techniques that achieve up to 7x faster prefill speeds and 500 tokens per second generation while previewing a simplified vLLM-style Python API coming soon.

about 9 hours ago · 10 points
March 10 - Jetson AI Lab Research Group Call - Lightning talks
55:28
NVIDIA AI Podcast NVIDIA AI Podcast

March 10 - Jetson AI Lab Research Group Call - Lightning talks

This Jetson AI Lab Research Group call features lightning talks on open-source hardware for remote Jetson access, a real-time emotional AI engine for robots running entirely on Jetson Nano, and updates to the Jetson AI Lab model repository with new performance benchmarks and deployment guides.

about 9 hours ago · 8 points
Feb 10 - Jetson AI Lab Research Group Call - Drones on Jetson & Isaac Lab on DGX Spark
57:34
NVIDIA AI Podcast NVIDIA AI Podcast

Feb 10 - Jetson AI Lab Research Group Call - Drones on Jetson & Isaac Lab on DGX Spark

Cameron Rose presents 'Operation Squirrel,' an autonomous drone project using Jetson Orin Nano for real-time target tracking and dynamic payload delivery. The system uses a modular C++ software stack with TensorRT-optimized YOLO and OSNet running at 21 FPS, communicating via UART with a flight controller to maintain following distance through velocity commands.

about 9 hours ago · 9 points