01.0 Setup of NVIDIA Jetson AGX Xavier
Our setup includes the Jetson AGX, two D455 cameras, one IMU from Genesys (ADMA Slim) as well as several CAN-Sensors and actors (for example two motors, inverters, wheelspeed sensors, BMS, ...) as can be seen in the image below.
- NVIDIA Jetson AGX Xavier with 32 GB RAM, upgraded 512 GB SSD and upgraded Intel Wifi 8265
- Intel D455 Cameras (x2)
- Genesys Adma Slim IMU
- Some kind of CAN Transceiver, DB9 connectors, cables, pins and other basic electrical hardware
- Ubuntu 18.04 LTS
- Python 3.X
- ROS 1 Melodic
- JetPack 4.4
On the following pages you will find a detailed description how to setup our system communicating manly via CAN-bus (sensors and actors) and USB3.0 (cameras).
You will need a host pc with Ubuntu 18.04 (or 20.04 with a little hack described here) in order to flash the AGX. First, install the NVIDIA SDK Manager on the host pc. Now, create a NVIDIA account if you don't have one already. Login and run the SDK Manager (Link). Refer to the manual inside the AGX delivery box on how to set it into a flashable state. Make sure to add the Cuda libs (and others you might need) via SSH after flashing the AGX with the Jetpack software.
Step 1:
Step 2:
Now that the basic Jetpack is set up on the AGX we need to install all custom libraries we will use in order to get the system running. This process is described on the following pages:
- Intel Kernel Patch and Open CV for AGX
- Library Installations for Python and ROS
- Tools and IDE
- ROS Melodic Installation
- Setup CAN Communication on NVIDIA Jetson AGX Xavier
When the installation described on these pages worked accordingly, you are now all set for the actual workspace setup as described here:
