wengkd/jetson repository overview

⁠NVIDIA Jetson Image

This repository provides optimized Docker images for NVIDIA Jetson devices, integrating machine learning (ML) and ROS 2 Humble for robotic and AI applications.

These images are built on top of NVIDIA's official Jetson images, ensuring compatibility with JetPack SDK and hardware acceleration.

⁠Base Images

These images are based on the following NVIDIA container sources:

• dustynv/l4t-ml⁠: Includes TensorRT, PyTorch, TensorFlow, and other ML libraries for Jetson.
• nvcr.io/nvidia⁠: NVIDIA’s official Jetson machine learning container with GPU-accelerated deep learning frameworks.

⁠How to Use

⁠Pull an Image

Choose the appropriate image tag based on your JetPack version:

docker pull wengkd/jetson:r35.4-torch-ros2

⁠Run a Container

docker run --runtime=nvidia --network=host -it --rm wengkd/jetson:r35.4-torch-ros2

• --runtime=nvidia → Enables GPU acceleration on Jetson.
• --network=host → Allows ROS 2 to communicate properly.
• -it --rm → Starts an interactive shell and removes the container on exit.

⁠Activate ROS 2

Inside the running container:

source /opt/ros/humble/install/setup.bash

Now you can start using ROS 2 Humble.

⁠Additional Dependencies for OpenCV Compatibility

To ensure cv2 works correctly, you may need to install the following packages:

sudo apt-get install -y libopencv-calib3d-dev
pip install opencv-contrib-python