ONTrack — Field Route Optimizer

Route optimizer for field technicians and service/delivery crews. Import a list of stops, get back an efficient driving order, and open it in your maps app of choice.

Ships as two native apps sharing one core library:

• Desktop — a three-panel egui GUI for Linux, Windows, and macOS.
• Android — a NativeActivity app with a Slint UI, distributed via Google Play and F-Droid.

---

Preview

The Android app, recorded from the current build: adding stops on Home, the Results page, and Settings.

Watch the full video preview (MP4)

The desktop app, recorded from the current build: adding the same five stops on Home and optimizing the route.

Watch the desktop preview (MP4)

---

Quick Start

Get ONTrack from the GitHub Releases page — the latest release is v2026-09-10.

• Android phone — get the app from the releases page.
• Windows desktop — download ontrack-windows-x86_64.zip from the v2026-09-10 release.

---

Developer info

git clone https://github.com/qompassai/ontrack.git
cd ontrack

# Build and run the desktop GUI
cargo build --release -p ontrack-desktop
./target/release/ontrack

# Run the core library's tests
cargo test -p ontrack-core

Crates

Crate	Type	Description
ontrack-core	library	Address parsing (CSV/XLSX), geocoding (Nominatim/Google), distance matrix (haversine/OSRM/Google), nearest-neighbor + 2-opt TSP solver, export URL builders, optional on-device Whisper voice input
ontrack-desktop	binary (ontrack)	egui desktop GUI — Linux, Windows, macOS
ontrack-mobile	cdylib + Android app	NativeActivity + Slint UI, built with cargo-ndk and packaged with Gradle

Solver

Nearest-neighbor seed + 2-opt local search — pure Rust, no C++ FFI, no OR-Tools. For typical field routes (≤ 50 stops) this produces near-optimal results in well under a second.

Backend	Algorithm	Quality	Speed (50 stops)
2-opt (default)	NN seed + 2-opt local search	Near-optimal	< 1s
Nearest-neighbor	Greedy NN	Good	< 10ms

Architecture

ontrack/
├── Cargo.toml                          # workspace
├── crates/
│   ├── ontrack-core/
│   │   └── src/
│   │       ├── lib.rs
│   │       ├── config.rs               # .env / environment loading
│   │       ├── parser.rs               # CSV / XLSX address files
│   │       ├── geocoder.rs             # Nominatim + Google geocoding
│   │       ├── matrix.rs               # haversine + OSRM + Google distance matrix
│   │       ├── solver.rs               # nearest-neighbor + 2-opt TSP
│   │       ├── exporter.rs             # CSV, Maps URL, FieldMaps, Street View, Waze
│   │       └── voice.rs                # optional on-device Whisper voice input
│   ├── ontrack-desktop/
│   │   └── src/
│   │       ├── main.rs
│   │       ├── app.rs                  # worker-thread app state
│   │       └── views/                  # home / results / settings panels
│   └── ontrack-mobile/
│       ├── src/
│       │   ├── lib.rs                  # android_main entry point
│       │   ├── controller.rs           # Slint UI wiring
│       │   ├── gps.rs                  # Android LocationManager via JNI
│       │   └── preview_main.rs         # desktop preview of the mobile UI
│       ├── ui/app.slint
│       └── android/                    # Gradle project (Play Store + F-Droid)
├── scripts/                            # build, sign, emulate, screenshot tooling
├── fastlane/                           # Play Store release automation
├── fdroiddata/                         # F-Droid metadata recipe
└── playstore/                          # Play Console listing metadata

---

Configuration

Copy .env.example to .env in the repo root:

cp .env.example .env

GOOGLE_MAPS_API_KEY=""    # optional — enables Google geocoding/distance matrix + Street View
OSRM_BASE_URL="http://router.project-osrm.org"
ARCGIS_ITEM_ID=""         # optional — your ArcGIS Online web map ID, for FieldMaps deep links
ONTRACK_WHISPER_MODEL="base"

Getting the keys

GOOGLE_MAPS_API_KEY — optional; only needed if you want Google for geocoding and driving distances instead of the free backends below.

1. In the Google Cloud Console, create or select a project.
2. Enable the Geocoding API and the Distance Matrix API for that project.
3. Go to APIs & Services → Credentials → Create credentials → API key.
4. Recommended: restrict the key to those two APIs, so it cannot be used for anything else.

Full walkthrough: Google’s Get API key guide.

ARCGIS_ITEM_ID — not an API key. It is the ID of your own ArcGIS Online web map, used to build Field Maps deep links.

1. Sign in to ArcGIS Online (a free account works) and create or open your web map.
2. Open the map’s item details page.
3. The item ID is the id= value in the page URL — copy it into .env.

See the ArcGIS Online item details documentation.

Free backends — no key at all. Nominatim geocoding and the public OSRM router (OSRM_BASE_URL) work without an account. If you send heavy traffic, respect the Nominatim usage policy or self-host OSRM instead.

ONTRACK_WHISPER_MODEL — picks the on-device Whisper model for voice input (for example base). No key and no account.

No API key required. Without one, the app uses:

• Nominatim (OpenStreetMap) for geocoding
• The public OSRM router for real driving distances, or offline haversine distance
• Plain Google Maps / Waze / Apple Maps URL schemes for turn-by-turn navigation

---

Android build (Play Store / F-Droid)

The Android app is built in two stages: cargo-ndk compiles the Rust core into libontrack_mobile.so for arm64-v8a and armeabi-v7a, then Gradle packages it into an APK/AAB. Nothing is committed as a prebuilt binary — both stores build from source.

# One-shot build (AAB for Play Store, APK for sideloading/F-Droid, or both)
./scripts/build-android.sh both

# Local emulator smoke test (SDK setup, AVD, install, launch, screenshot)
./scripts/acli.sh full ontrack 4

See scripts/ for the full Android/emulator/signing toolchain, and fastlane/ + fdroiddata/ for the store metadata.

Application ID: ai.qompass.ontrack

---

License

Apache License 2.0 — see LICENSE.md.