Installation & Prerequisites

Before starting your data collection pipeline, assume your environment has been successfully built according to the ROS 2 Compilation guide.

The primary dependencies should already be covered natively:

• GStreamer (gi): Core video extraction and formatting logic.

• opencv-python: Utilities consumed structurally by the extraction arrays.

• rclcpp / rosbag2_cpp: The backend logging frameworks powering the ROS 2 decoupled architecture.

NVIDIA Hardware Acceleration

To use NVIDIA GPUs for hardware-accelerated video encoding (NVENC), you need to have the NVIDIA drivers and the GStreamer nvcodec plugin installed. This is often provided by the NVIDIA DeepStream SDK.

sudo apt update
sudo apt install libnvidia-encode-*

To verify availability, run:

gst-inspect-1.0 nvh264enc

If you just installed DeepStream or new GStreamer plugins and they are not appearing in gst-inspect-1.0, you may need to clear the GStreamer registry cache:

rm -rf ~/.cache/gstreamer-1.0

Installation & Environment

Before running the application, ensure your ROS2 environment is sourced:

source /opt/ros/YOUR_DISTRO/setup.bash
# or if using a workspace
source install/setup.bash

To build the workspace:

colcon build --symlink-install