AD-R1M System Setup
Base AD-R1M
Todo
Flash SD card image
Connect to WiFi and update AD-R1M software
AD-R1M + Nvidia AGX Orin
This section describes the main steps required to flash an NVMe with JetPack 6.2.1 on the NVIDIA AGX Orin and to set up the environment for using Isaac ROS packages (cuVSLAM in this case). It also covers configuring RealSense cameras using an NVIDIA-provided Docker image and integrating cuVSLAM with the AGX Orin and AD-R1M navigation functionalities (Sections 1-7).
- 1) Linux installation on Jetson AGX Orin
- 2) Setup required packages on AGX Orin
- 3) Setup Isaac ROS on AGX Orin
- 3.1) Set locale
- 3.2) Add the universe repository
- 3.3) Setup ROS2 source
- 3.4) Install ROS2 packages
- 3.5) Check apt sources list
- 3.6) Configure nvidia-container-toolkit
- 3.7) Install Large File Storage for git
- 3.8) Create a Isaac ROS workspace
- 3.9) Setup the Isaac ROS workspace
- 3.10) Build the Isaac ROS workspace
- 3.11) Cleanup
- 4) Setup Isaac ROS Visual SLAM
- 5) VSLAM and RealSense setup and usage on AGX Orin
- 6) AD-R1M and Jetson AGX Orin: hardware setup
- 7) AD-R1M and cuVSLAM setup
Below are the necessary steps to configure the AD-R1M to use visual odometry from cuVSLAM in its state estimation and to build a NAV2-compliant map for autonomous navigation using the landmarks point cloud provided by cuVSLAM(Sections 8–9).
If you want to test this setup in a more controlled environment, we recommend using our Gazebo simulation, which includes the base AD-R1M robot (equipped with an IMU, ToF camera, and wheel encoders) plus an additional RealSense camera (Section 10).
Prototype AD-R1M (Legacy)
Todo
Setup instructions for nonstandard development samples? Shouldn’t this just be internal-only?