Back to flagship work

Case study · 2026

Eco Route

An early-beta Android route planner that compares extra travel time with estimated fuel and cost savings on an interactive graph, with calculations running on your phone.

Download APK

v0.14.0 · Android 8.0+ · 6.42 MB

My role: Independent product designer and Android engineer; route-model implementation

Outcome

Built a map-first planner with a selectable time–fuel graph, foreground driving guidance, multiple stops, return trips, and saved cars. Fuel predictions remain uncalibrated estimates.

Latest signed phone release

Download Eco Route

Download APK

Free and open source. Early beta for Android. No account, advertising, or in-app analytics. New address searches, roads, map tiles, and fuel prices require Internet; calculations and saved cars stay on your phone.

Release
v0.14.0 · code 19
Compatibility
Android 8.0+ · phone APK
Release details (checksum, package)
Package
org.kalc.ecoroute
File size
6.42 MB (6,420,457 bytes)
SHA-256
47728cfb43c5a8513333dcb5482612a74e4e82a17bfd2120d2eb5901e66aa065

This APK installs outside Google Play. Android may ask you to allow installs from your browser. Verify the checksum above before installing, and only grant that permission for the installation.

Eco Route 0.14.0 time–fuel graph showing extra travel minutes and estimated litres saved with synthetic test data
v0.14.0: choose extra minutes and estimated fuel savings; synthetic test data
Eco Route offline car search displaying Volvo XC90 reference entries
Offline vehicle references and saved cars; captured in v0.9.0
Android early-beta screens. Graph values use synthetic test data and are not measured savings.

Fuel and travel time as a visible trade-off

The fastest route is not necessarily the route with the lowest fuel use. Eco Route plots actual extra travel minutes against estimated litres or euros saved, compared with the fastest modeled whole journey. Each dot selects routes and speed profiles across all stops and the return leg. Fastest and Most fuel saved shortcuts select the extremes; Suggested balance appears when a useful interior choice is detected among the sampled plans.

The destination-first interface keeps the map central. Address suggestions, recent places, an optional current-location start, multiple stops, and a return journey feed the same planner. Foreground driving mode follows the selected speed profile and clears its advisory when GPS is stale or the position is uncertain or off-route. It pauses while hidden. Follow posted signs and actual conditions; the app does not provide turn-by-turn navigation.

Public roads, private calculations

The native Kotlin and Jetpack Compose app takes public map data and does the calculation on the phone. No connection to my private server or Tailscale is required.

  • From public providers: addresses from Photon, road candidates and elevation from Valhalla, map tiles from OpenStreetMap, and a dated Slovenian reference fuel price from GOV.SI.
  • On the phone: the road-load model and speed optimization, plus saved cars, trip load, and editable physical assumptions.
  • Recent places: up to 20 confirmed places stay on the device; remove them individually or clear them all.
  • Offline car search: a bundled catalogue of 39,717 grouped petrol/diesel references from the EEA 2024 registration dataset. It is a reference catalogue, not a verified match for every trim: mass and laboratory consumption have provenance, while aerodynamic drag and other missing quantities use explicitly estimated, editable assumptions.
  • Not covered: electric and plug-in vehicles are outside the current model.

There is no account, advertising, in-app analytics, or background location tracking. Public providers still receive address queries, route coordinates, or viewed map areas and ordinary connection metadata. Internet is needed for new online data. The privacy policy explains these requests, local caching, optional location access, exports, and deletion.

The model and its boundaries

The road-load-v2 calculation combines vehicle and trip mass, road grade, aerodynamic drag, rolling resistance, acceleration, drivetrain-efficiency assumptions, and auxiliary demand. A constrained search compares discrete speed profiles within the time budget on the supplied road candidates. It does not establish the globally optimal route across the entire road network.

Fuel values are uncalibrated estimates. The model does not simulate an engine map, hybrid battery state, or energy recovery. Live traffic, weather, time spent at stops, tolls, and parking charges are excluded. Cost uses a dated Slovenian outside-motorway reference fuel price even for a route abroad; it is not a quote from a station along the journey.

A shareable early beta

The download is the signed Android 0.14.0 phone APK (code 19), for Android 8.0 and later. Its permanent signing identity is retained so existing compatible installations can update. The release publishes the package identity, exact byte size, SHA-256, corresponding build source, and dependency notices.

The release checks covered three areas:

  • Model parity: numerical checks compared the Android model with its Python reference.
  • Core flows: automated and emulator checks exercised graph selection, minutes and euros, map and export consistency, and return trips.
  • Upgrade path: a signed upgrade retained cars, load, recent places, and preferences.

These checks establish software consistency and exercised flows, not measured fuel accuracy, savings, or comprehensive physical-device reliability. The public download is intentionally labeled early beta.

From addresses to a fuel estimate

  1. 01

    Choose your trip

    Select addresses, stops, and a saved car.

  2. 02

    Fetch road candidates

    Public providers supply roads, elevations, map tiles, and a reference price.

  3. 03

    Calculate on your phone

    Choose a graph dot to compare extra minutes, estimated fuel savings, and cost locally.

Simplified Android workflow. Results depend on road data and estimated vehicle assumptions; new online data needs Internet.

Technology focus

  • Kotlin
  • Jetpack Compose
  • Route Optimization
  • OpenStreetMap

Evidence and further reading

Evidence reviewed .

Back to top ↑