Darkstar Robotics · Native macOS · Managed Linux Runtime
v0.3.0 — ReleasedDarkstar Robotics is a robotics development workstation that lives on your MacBook while the Linux runtime runs managed underneath. No terminal. No apt locks. No version matrix. One calm place for ROS, Gazebo, PX4 — and the firmware inside the robot.
Everything a budding roboticist needs, arranged the way a flight line is: check the board, run the checks, pick the environment, load the mission.
Live state of the managed Linux runtime, the active Ubuntu / ROS / Gazebo pairing, PX4 SITL, and the Tailscale link — at a glance, without opening a terminal.
A prerequisite checklist that probes this Mac and the runtime before anything flies — each failed item gets a one-click remedy where one exists.
Validated pairings, not version roulette. Pick what you want to build — Darkstar Robotics already knows which Ubuntu, ROS, and Gazebo versions go together.
Photo and scan intake. Drop a picture of a board, flight controller, label, or QR code — your fleet starts from the hardware you actually own.
The PX4 swarm-recovery golden demo, shipped with its verified evidence — separation, samples, and reassignments on the record, not in a slide deck.
Embedded replay and telemetry viewing, so a flight can be re-flown from the couch and every anomaly inspected frame by frame.
A recorded walkthrough of the Darkstar Robotics app, driven page by page — Settings, Dashboard, Preflight, Environments, Hardware, Missions, Viewer — exactly as it runs on the Mac. Real frames of the installed v0.3.0 app, not a render. Press play.
Darkstar Robotics is not an all-or-nothing installation. Nobody should have to install ROS 2, Gazebo, PX4, Docker, and dozens of other tools just to open the application. You start with the Core — everything else is opt-in.
“Darkstar Robotics Core must remain independently usable. Robotics applications, runtimes, SDKs, and hardware integrations are opt-in. Readiness is evaluated against the user's selected capabilities, not the entire software catalog.”
Product principle — Phase 1 spec
The lightweight desktop application: system inspector, compatibility engine, application catalog, and diagnostics interface. No ROS, Gazebo, PX4, or simulator installation required to start.
Choose individual tools only when you need them — ROS 2, Gazebo, PlatformIO, OpenCV, QGroundControl, KiCad. Want only QGroundControl? Nothing forces ROS 2 or Gazebo along with it. Building for an ESP32? PX4 stays out of it.
Curated combinations for drone development, rovers, robot arms, embedded systems, and computer vision. Every component is individually selectable — a stack is a starting point, never an all-or-nothing bundle.
A selected tool may need other software to work. Darkstar Robotics explains those dependencies and asks for approval — it never installs anything silently. Every dependency in the catalog is one of three types:
Must be present for the selected capability to work at all. If it's missing, that capability — and only that capability — reports it.
Adds functionality, but the selected capability works without it. An absent optional dependency is never an error.
A choice between supported implementations. Pick the one your workflow already uses; the catalog tracks either.
Is Darkstar Robotics itself functioning correctly — the Core application, its system inspector, and its diagnostics?
Measured against the Core onlyDoes this machine meet the requirements of the specific tools and workflow you selected — and only the ones you selected?
Measured against your selectionNever selected PX4? Its absence is not an error and raises no dashboard warning. It simply reads Not Installed in the catalog — neutral, until the day you choose it.
CONCEPT LAYOUT — this page installs nothing and verifies nothing about your machine. Stack templates are proposed combinations; each needs a version matrix before Darkstar Robotics can claim it compatible. In the app, dependencies are explained and approved before anything installs.
Before setup, before flight, Darkstar Robotics probes the host, the runtime, and the environment. Failures say why — and fix themselves where they can. Try the sequence:
The OS version decides the ROS distro — 24.04 means Jazzy, 22.04 means Humble. Darkstar Robotics encodes that once, so no new user ever has to learn it the hard way.
The working stack today — it's the pairing running inside the Omega-Ubuntu prototype: Ubuntu 24.04.3 LTS, tuned to a 6.4-second boot with zero failed units, ROS 2 sourced and ready, PX4 simulation on top.
The next validated pairing, for the huge installed base still on 22.04 / Humble. Same rule as everywhere in Darkstar Robotics: you pick what you want to build — never which versions "go together."
Most robotics tooling starts with a shopping list. Darkstar Robotics starts with your bench. Photograph a board, a flight controller, a label, a QR code — the fleet is built from the hardware already in front of you.
It's the input method nobody else ships: no catalog to scroll, no part numbers to memorize.
The golden demo isn't a render. It's a PX4 swarm-recovery flight with its numbers attached — and every status on this page is labeled for what it actually is: live, recorded, or pending. Green means verified.
Darkstar Robotics v0.3.0 is out. Download the installer, open it, and it walks the app into place — logo, status check, and all.
The installer GUI checks for an existing install, shows what it found, and installs Darkstar Robotics to your Applications folder in one click. Prefer to skip the installer? The app-only zip is the same build, ready to drag in yourself.
v0.1 is public and source-available. Clone it, open the package, and the workstation is yours.
# Clone Darkstar Robotics $ git clone https://github.com/abelxmendoza/HangarBay.git $ cd HangarBay # Run it — or just open Package.swift in Xcode $ swift run # Settings (⌘,) holds the VM name + SSH target. # → Dashboard live. Runtime probing…
