# Isaac ROS Nav2 Demo — NVIDIA Jetson The Isaac ROS Nav2 demo integrates NVIDIA Isaac ROS (GPU-accelerated ROS 2 packages) with the ROS 2 Nav2 navigation stack on Jetson hardware. It demonstrates autonomous navigation with GPU-accelerated perception — replacing the CPU-bound costmap update pipeline with cuVSLAM (GPU visual SLAM) and Isaac ROS Nvblox (GPU 3D scene reconstruction). This demo targets real or simulated differential-drive robots and provides end-to-end autonomy: sensor data → GPU-accelerated costmap → Nav2 planner → velocity commands. ## Hardware - NVIDIA Jetson Orin (AGX, NX, or Nano) — Orin recommended for best GPU performance - Intel RealSense D435i or similar depth camera (visual SLAM input) - Differential-drive robot base (e.g. iRobot Create 3, TurtleBot 4) - External LIDAR (optional — Nvblox can replace with depth) ## Key Packages - `isaac_ros_visual_slam` — cuVSLAM odometry - `isaac_ros_nvblox` — GPU 3D occupancy mapping - `nav2_bringup` — path planning, costmaps, recovery behaviors - `isaac_ros_nitros` — zero-copy tensor transport between GPU nodes ## Source - Repo: https://github.com/NVIDIA-ISAAC-ROS/isaac_ros_nvblox - License: Apache 2.0 - Docs: https://nvidia-isaac-ros.github.io/
Category: Other Robots
| Item | Qty | Notes | |------|-----|-------| | NVIDIA Jetson Orin AGX/NX/Nano | 1 | Jetson Orin recommended for nvblox GPU perf | | Intel RealSense D435i | 1 | Depth + IMU for cuVSLAM | | TurtleBot 4 or iRobot Create 3 | 1 | Differential-drive base | | External LIDAR (optional) | 1 | Augments Nvblox occupancy map |