In robotics development, effective visualization and analysis tools are essential for monitoring, debugging, and interpreting complex sensor data. Platforms like RViz, Foxglove, and Rerun play a key role at the visualization layer of the observability stack. They help developers interact with both live and recorded data. These tools rely on timely, well-structured access to the underlying data streams. That's where ReductStore comes in. It handles the data logging, storage, and processing, with a focus on capturing high-volume time-series data efficiently. ReductStore aims to integrate with tools like RViz, Foxglove, and Rerun, supporting a complete observability pipeline: from raw data ingestion to actionable insights.

When building computer vision systems, efficient data storage is a fundamental requirement. Whether you're capturing images for training, storing inference results for validation, or archiving sensor data for future analysis, your storage solution must be both reliable and high-performance.

Ingestion speed is especially critical. If your system can’t write data fast enough — whether it’s high-frequency frames or accompanying metadata — you risk losing valuable information or creating bottlenecks in the pipeline.

In this post, we’ll look at three common approaches to storing data in computer vision applications: a traditional file system, S3-compatible object storage, and ReductStore, a time-series-optimized blob storage. We’ll explore the strengths and limitations of each approach to help you choose the best fit for your application.

A Simple Computer Vision Application​

For demonstration purposes, we’ll use a simple computer vision (CV) application which is connected to a CV camera and runs on an edge device:

The camera driver captures images from the CV camera every second and forwards them to the model, which then detects objects and displays the results in the user interface.

Your images and results need to be stored for training and validation purposes. The customer may also wish to view images featuring anomalous objects. These requirements present the challenge of maintaining a history of blob or unstructured data.

At ReductStore, we specialize in the high-performance storage and streaming of robotics data from edge devices to the cloud. In this tutorial, we will demonstrate how to develop a robust ROS 2 data logging pipeline for practical robotics applications.

First, we will set up a Raspberry Pi with a USB camera running a lightweight YOLOv5n object detection model via ONNX Runtime. Then, a recorder node will capture selected ROS 2 topics, including images, detection results, and logs. Next, these topics will be saved as segmented MCAP files locally with ReductStore. Finally, we will configure automatic replication to stream data to another ReductStore instance.

This minimal setup shows how to efficiently capture, store, and replicate ROS 2 data from a robot to a central server or cloud instance. These techniques can be applied to any ROS 2 system, whether it's a single robot or a fleet of autonomous systems.

Let's get started!