# Design a Mobile Astronomy / Sky Map AR App

Source: https://www.techinterview.org/post/3233475380/mobile-system-design-astronomy-sky-map-ar/
Updated: 2026-05-06 · techinterview.org

"Design a sky map app" is a sensor-and-AR mobile system design prompt — Sky Guide, SkySafari, Stellarium, Star Walk are the references. The interview tests whether you understand sensor fusion (gyro + accel + magnetometer), real-time AR overlay, astronomical computation, and the unique requirements of a "point at sky → identify object" experience.

## Clarify scope

- Stars only or also planets, satellites, comets?

- AR overlay or static sky map?

- Predictive ("when will Saturn rise") features?

- Offline operation expected?

- Telescope control (ASCOM, smart-telescope APIs)?

## The sensor fusion problem

Pointing the phone at the sky requires knowing:

- Where the phone is (GPS, lat/lon)

- What time it is (system clock, validated against time server)

- Which way the phone is pointing (sensor fusion of gyro, accel, magnetometer)

- Calibration of the magnetometer (the hardest part)

## Sensor fusion in practice

- iOS: CoreMotion provides device attitude (CMAttitude) — already fused

- Android: SensorManager.getRotationMatrix combines accel + magnetometer; or use ROTATION_VECTOR sensor

- Drift handling: the magnetometer drifts; periodically calibrate against gyro+accel

- "Point at sun and tap" calibration (Stellarium-style) for indoor / interference scenarios

## Astronomical computation

- Star catalog: ~100K stars (Hipparcos, Tycho-2) embedded in app (~50MB)

- Planet positions: VSOP87 or simplified Kepler — computed at runtime

- Moon: ELP-2000 or similar; high accuracy required (close to user)

- Constellations: hand-curated line patterns connecting bright stars

- Deep-sky objects (galaxies, nebulae): ~10K from NGC catalog

- Satellites: TLEs (two-line elements) updated weekly via celestrak.org

- Comets: orbital elements; less common

## Ephemeris freshness

- Star positions: stable for human lifetimes (proper motion negligible at amateur scales)

- Planets: computed; no update needed

- Satellites: orbital elements decay; update TLEs every 1–7 days

- Comets: update on outburst or new discovery

## The render pipeline

- Convert observer location + time → local sidereal time

- For each visible object: convert (RA, Dec) → Alt-Az coordinates

- Filter to objects above horizon

- Project to phone screen using device attitude

- Render with Metal / OpenGL ES / Vulkan

## AR overlay

- Camera preview as background

- Star labels and constellation lines overlaid using sensor-derived attitude

- Tap an object to identify; see info card with distance, magnitude, history

- "Tonight's highlights" auto-suggestions

## Offline operation

- Star catalog and planet algorithms ship with the app

- GPS works offline (just slower without A-GPS)

- TLEs are the only thing requiring periodic update

- App should fully function in remote stargazing locations with no signal

## Battery considerations

- Camera + sensors at 30 Hz is heavy

- Drop sensor poll rate when device idle (no movement)

- Allow night mode (red display) to preserve dark adaptation

- Auto-dim when not interacting

## Night mode

- Red-tinted UI to preserve dark adaptation

- Reduced brightness override

- Toggle prominent in main view

## Educational features

- Constellation stories (mythology)

- "Tonight" page with what is visible from current location

- Planning tools: ISS pass times, conjunction events, eclipses

- Notification: "ISS visible in 5 minutes from your location"

## Telescope integration

- Smart telescopes (Vespera, Stellina, ZWO Seestar) connect via Wi-Fi

- App can drive go-to commands from the AR view

- Live image streaming from telescope camera

- Out of scope for most apps but a niche premium feature

## Performance considerations

- Stars rendered as point sprites; 100K is fine on modern GPUs

- Use depth-sort only for transparency (planets behind/in-front)

- Frame rate target: 30+ fps for AR overlay

- Memory: ephemeris computations are cheap; the catalog dominates RAM

## Astronomical accuracy

Amateur apps target arc-minute accuracy:

- Stars: better than 1 arcsec (Hipparcos catalog)

- Planets: arcsec-level with VSOP87

- Moon: arcmin-level

- Sun: arcsec-level

- Satellites: km-level position; arcmin-level apparent

Professional astronomy uses higher-fidelity algorithms; not needed for consumer apps.

## What separates senior from staff

Senior candidates handle sensor fusion and basic rendering. Staff candidates address magnetometer drift, the calibration UX, and the offline ephemeris story. Principal candidates discuss telescope-control protocols, the educational content pipeline, and the very-large-aperture (computational photography for astrophotography) extension.

## Frequently Asked Questions

### How do I handle indoor / shielded magnetometer?

Detect with the magnetometer accuracy reading; prompt user for "calibration dance" or manual point-at-known-object calibration. Most apps degrade gracefully but get inaccurate.

### Where do the catalogs come from?

Hipparcos, Tycho-2, Gaia DR3 (now standard) for stars. NGC + IC for deep-sky. JPL Horizons for planet/comet ephemerides. All public-domain or permissive-license.

### What about light pollution overlay?

Some apps overlay light-pollution maps from VIIRS satellite data. Useful for stargazing-trip planning. Out of core but a frequent enhancement.