Robopipe OS is the operating system for Robopipe controllers, the industrial edge computers (AI PLCs) that run AI visual quality inspection on production lines. This repository contains the build scripts and the ready-to-install OS images.
Robopipe OS is part of Robopipe, an industrial machine vision platform for AI quality control in manufacturing. The image is based on the Unipi Patron base OS (Debian) and comes with Robopipe API preinstalled. After installation, the controller connects Luxonis OAK cameras to the PLC inputs and outputs and runs AI inference at the edge.
- Robopipe API preinstalled. The build installs the latest Robopipe API release as a system service, with nginx in front of it.
- Ready for the network. Each controller gets the hostname
robopipe-<id>.localand announces itself through mDNS (Avahi). - PLC inputs and outputs. The image contains the Unipi hardware definitions, so Robopipe API can control digital, analog and relay inputs and outputs.
- 1-Wire support. The image contains owserver for 1-Wire sensors.
- Software update in the browser. Install a new OS version through the service mode web page of the controller.
- Reproducible builds. Build the image yourself in Docker. GitHub Actions publishes each tagged version as a release.
| Part | What it does | Where it runs |
|---|---|---|
| Robopipe Studio | Capture, labelling, training and deployment of AI models | Server or cloud, used in a web browser |
| Robopipe API | REST and WebSocket API for cameras, AI inference and PLC inputs and outputs | Robopipe controller or any Unix-based computer |
| Robopipe OS (this repository) | Operating system image with Robopipe API preinstalled | Robopipe controller (based on Unipi Patron) |
| Luxonis OAK cameras | Image capture and on-device neural network inference | Above the production line |
Read the Robopipe documentation for a detailed guide.
- Start the controller in service mode:
- Turn off the controller.
- Press and hold the SERVICE button. It is usually next to the USB labels.
- Turn on the controller and continue to hold the SERVICE button.
- Release the SERVICE button when all LEDs blink periodically.
- Download the archive.swu file from the latest release of Robopipe OS.
- Enter the IP address of the controller in your browser to open the service mode web page. The documentation shows how to find the IP address.
- Drag the archive.swu file into the Software Update window, or click the window and select the file.
The update takes a few minutes. After the restart, open http://robopipe-<id>.local. The page shows Hello from Robopipe API!.
Prerequisites: Docker with support for linux/arm64 (QEMU on x86 computers).
git clone https://github.com/Robopipe/OS.git
cd OS
./build-in-docker.shThe script builds archive.swu in the repository folder. It downloads the Unipi Patron base OS and the latest Robopipe API release. To use a local base image, put it at data/base-os.zip. The files in data/ contain the controller configuration, the Robopipe API settings and the systemd services.
To publish a release, push a tag. The GitHub Actions workflow builds the image and attaches archive.swu to the release.
Robopipe is an alternative to Roboflow, Cognex and Keyence for AI visual inspection on production lines. The Robopipe controller with Robopipe OS connects the camera and the AI model to the line through PLC inputs and outputs. Read the comparisons: Roboflow alternatives, Cognex alternatives and Keyence alternatives.
- Food processing: defects, foreign objects and portion checks
- Pharma and healthcare: fill, closure and label verification
- Retail and e-commerce: automated product tagging
- Logistics: safety hazards and movement statistics
See the case studies and the Robopipe blog.
If you have a problem with the default configuration of the OS or a suggestion, open a new issue.
Robopipe OS is open source under the MIT License. The OS image also includes the Unipi Patron base OS and third-party packages under their own licences.
- Website: robopipe.io (česky)
- Book a demo or ask a question: robopipe.io/en/contact
Robopipe is developed by Robopipe s.r.o., Prague, Czech Republic.