After 6+ years in robotics, the pattern that kept frustrating me was field debugging. When a deployed robot stops doing what it should, whoever is standing next to it is often stuck with a laptop on one knee, several terminals open, ros2 topic echo scrolling past, and logs to tail across a dozen nodes, services, and processes, all while trying to find the cause before it becomes a customer problem.
So I built Scry: an AI-first debugging tool for ROS 2 robots that runs on your phone.
How it works
Scry is two pieces:
- scry-connect — a small Python server you run on the robot (
pip install scry-connect). It usesrclpy, so it’s RMW/middleware-agnostic and works with Fast-DDS, CycloneDDS, Zenoh, and Connext. It exposes the robot’s ROS 2 graph to the phone over your own network (no cloud). - The Scry Android app — the thick client. It runs the AI loop, renders results, and talks to the robot.
You ask a question in plain English — “why isn’t the robot moving?” — and the AI inspects topics, nodes, services, parameters, lifecycle, TF, and logs live, then reports the likely root cause. Reads are free; anything that changes the robot (publishing, setting a parameter, calling a service, lifecycle changes) requires explicit on-screen approval first.
What it can do
- Browse every ROS 2 entity from your phone: topics, nodes, services, actions, lifecycle, parameters, components, TF frames, logs, and processes — each as a searchable list with one-tap detail.
- Live visualizations: camera, LiDAR, plots, behavior trees, and a top-down 3D scene.
- Background monitors (e.g. “alert me if
/odomdrops below 10 Hz”). - Multi-robot fleet view.
AI providers
Bring an OpenRouter key for cloud inference (300+ models including Claude, GPT, Gemini, Llama, DeepSeek), or run fully offline against a local Ollama server. Your key, your robot, your network — no telemetry, no cloud backend.
Supported ROS 2 distros: humble, iron, jazzy, kilted, lyrical, rolling.
It’s still early and I’d really value feedback from this community — especially on the tool coverage and where it falls short on real robots.
- App (Google Play): https://play.google.com/store/apps/details?id=com.phaneronrobotics.scry
- Robot server (PyPI): scry-connect · PyPI
- Docs: Scry — debug your ROS 2 robot from your phone - Scry
- Demos: Scry demos — see it in action - Scry
- Video playlist: Scry — Demos - YouTube
- GitHub: GitHub - phaneron-robotics/scry: ros2 ros robotics mcp model-context-protocol android ai llm debugging rclpy jetpack-compose kotlin python developer-tools robot · GitHub