DC data collection framework updated for ROS 2 Jazzy, Lyrical, and Rolling

The DC (Data Collection) framework for ROS 2 has received a significant update, expanding its compatibility to include the Lyrical, Jazzy, and Rolling distributions. Designed to run directly on robotic hardware, the tool automates the collection of measurements and validates them against JSON schemas before transmission.

This update introduces a new uploader capable of moving heavy data types, such as camera frames, maps, and videos, to object storage using verified and resumable transfers. Additionally, the project has transitioned from Fluent bit to Vector to simplify integration and improve data pipeline management.

Technical improvements in DC 0.2

• Data Validation and Storage: The framework validates every collected record against a JSON schema and supports shipping data to PostgreSQL databases or object storage.
• Enhanced File Handling: A new uploader facilitates the transfer of large files like video and maps to object storage, ensuring metadata for every file is stored as a queryable record.
• Deployment and Containerization: The project now provides maintained container images on GitHub Container Registry (ghcr.io) for each supported distro, alongside deployment files for Docker Compose, Quadlet, and Kubernetes.
• Pipeline Refactoring: Support for Fluent bit has been removed in favor of Vector to streamline the integration process for edge deployment.

Why it matters

By providing structured, SQL-queryable records in PostgreSQL, DC allows robotics engineers to perform fleet-wide analysis and join robotic telemetry with other organizational data. The addition of Kubernetes and Quadlet support makes it easier to deploy these data pipelines in production-grade edge computing environments.