AccessU is an Android app that helps blind and low‑vision students navigate university campuses independently. It combines voice interaction, indoor QR routes, outdoor maps, and obstacle detection into a single, hands‑free experience.

Traditional campus maps and signage assume you can see. Even with mobility training, visually impaired students often need sighted assistance for:

• Finding the right building entrance
• Navigating complex indoor corridors, doors, and stairs
• Moving safely around crowds and obstacles

AccessU aims to reduce that dependence and increase confidence and independence on campus.

• Voice‑driven navigation

  • You speak your destination (e.g. “ETLC to NREF” or “Cameron Library”).
  • The app uses speech recognition to understand you and text‑to‑speech to guide you.

• Indoor navigation with QR codes

  • Buildings are modeled as graphs; each QR is a node (checkpoint).
  • You scan QRs along the route; AccessU:
    • Locates you in the graph
    • Runs a shortest‑path (BFS) to the destination
    • Speaks clear, step‑by‑step directions
  • If you scan a wrong‑path QR, it reroutes automatically.

• Outdoor navigation

  • Uses Google Maps and GPS to get you to the right building or bus stop.
  • Turn‑by‑turn style spoken cues (no need to look at the screen).

• Obstacle awareness

  • Camera feed is analyzed with ML Kit Object Detection.
  • Detects obstacles in a “path zone” in front of the user.
  • Speaks lightweight cues like “Obstacle ahead, move slightly left/right” or “Path is clear.”

• Fully voice‑first experience

  • No visual UI required during navigation.
  • Designed for screen‑free use: simple phrases, minimal cognitive load.

• Path graphs in JSON

  • Each indoor route (e.g. ccis_to_cameron, etlc_to_nref) is a JSON file:
    • Nodes: id, name, instruction, nextNodeId
    • nextNodeId == null marks the destination.
  • Graphs are stored in app/src/main/assets/paths/.

• Shortest path & rerouting

  • When a QR is scanned, payload is pathId:nodeId.
  • The app loads the corresponding graph via PathRepository.
  • A BFS over the adjacency list finds the path from current node to destination.
  • If you jump onto a wrong node, BFS recomputes the path from there (reroute).

• Spoken guidance

  • For regular checkpoints:
    • “On right path. {action instruction}”
  • For first and last checkpoints:
    • Just the full instruction (more detailed), no “On right path.”
  • Instructions are authored in the graph JSON, so they can be updated without code changes.

• Campus graph

  • Campus building graph is stored in campus/buildings.json.
  • Indoor path references (indoorPaths) link origin/destination building IDs to indoor path IDs.

• Google Maps + Location

  • Uses Google Maps SDK and Play Services Location to:
    • Get user location
    • Render map and geometry
    • Build outdoor routes between campus nodes

• Camera pipeline

  • A Compose screen (ObstacleCameraScreen) binds a CameraX pipeline.
  • Frames are passed to ObstacleDetector.

• ML Kit Object Detection

  • Uses com.google.mlkit:object-detection with:
    • ObjectDetectorOptions.SINGLE_IMAGE_MODE
    • enableMultipleObjects()
  • For each frame:
    • Filter detections by size and position inside a “center path” region.
    • Pick the closest relevant object.
    • Speak:
      • “Obstacle ahead, move slightly left.” or
      • “Obstacle ahead, move slightly right.” or
      • “Path is clear.”
  • Includes a global cooldown (~3.5 seconds) to avoid TTS spam.

• Future depth

  • TensorFlow Lite dependency is included for a future depth or semantic segmentation model.

• Platform & language

  • Android (minSdk 26, targetSdk 36)
  • Kotlin, Gradle (Kotlin DSL)

• UI

  • Jetpack Compose (Material 3)
  • Custom U of A–themed color system (UofAGreen, UofAGold, etc.)

• Camera & vision

  • CameraX (camera2, lifecycle, view)
  • Google ML Kit:
    • Object Detection (obstacle detection)
    • Barcode Scanning (QR codes for indoor navigation)
  • TensorFlow Lite (reserved for future depth/object models)

• Navigation

  • Google Maps SDK for Android
  • Google Play Services Location (GPS)
  • Custom BFS routing for:
    • Indoor path graphs (QR checkpoints)
    • Campus building graph

• Voice & accessibility

  • Android SpeechRecognizer (voice input)
  • Android TextToSpeech (spoken guidance)
  • Custom phrasing and timing (e.g. door pause: “Automatic door ahead.” then “You may continue.”)

• Data & infra

  • Gson for JSON parsing
  • Path and building data in local assets (offline‑friendly)
  • OkHttp (for future metadata / network features)

• Android patterns

  • Activity Result API (permissions, camera)
  • Compose navigation screens for flows (voice, indoor, outdoor, obstacle)

• app/src/main/java/com/example/accessu/navigation/
  High‑level nav flow, voice interaction, and screen routing.

• app/src/main/java/com/example/accessu/paths/
  Indoor path graphs, camera QR navigation, BFS routing.

• app/src/main/java/com/example/accessu/campus/
  Campus building graph, outdoor navigation, maps.

• app/src/main/java/com/example/accessu/obstacle/
  Obstacle camera pipeline, ML Kit object detection, spoken obstacle cues.

• app/src/main/assets/paths/
  Indoor route graph JSONs and metadata.

• app/src/main/assets/campus/
  Campus buildings graph and indoor path references.

• app/src/main/assets/qr_test/
  QR test pages to generate printable codes for indoor paths.

• Richer obstacle understanding

  • Integrate depth or semantic segmentation (via TensorFlow Lite) for:
    • Distinguishing people vs static obstacles vs walls
    • Better distance estimation and urgency

• Dynamic route generation

  • Generate indoor routes from building floor plans rather than fixed graphs.
  • Auto‑detect closures or blocked corridors and reroute.

• User customization

  • Adjustable verbosity of instructions
  • Different walking speeds and safety margins
  • Multi‑language support

• Crowdsourced corrections

  • Allow users and staff to suggest updates for QR placement and instructions.
  • Sync with a backend so buildings can be added and improved over time.

• More campuses and buildings

  • Expand beyond the initial routes (CCIS ↔ ETLC, ETLC ↔ NREF, bus stops, etc.).
  • Create a toolkit for other universities to deploy their own AccessU graphs.

• Advanced accessibility integration

  • Explore integration with white canes, haptic feedback, or wearable devices.
  • Better cooperation with TalkBack and other assistive services.

• Requirements

  • Android Studio (latest stable)
  • Android SDK 26+
  • A device or emulator with camera support (for QR + obstacles recommended on a real device)
  • local.properties with GOOGLE_MAPS_API_KEY

• Run

  • Open the project in Android Studio.
  • Add your Maps API key to local.properties:
    • GOOGLE_MAPS_API_KEY=your_key_here
  • Build and run the app module on a device.

Note: Indoor routes rely on printed QR codes placed in the environment; the qr_test HTML files help you generate them.

• Sanika Verma – Navigation flows, indoor/outdoor UX, voice experience
• Niharika Rawat – Obstacle detection, camera pipeline, accessibility behavior

Team RouteCause – overall concept, design, and testing.

Add your preferred license here (e.g. MIT, Apache 2.0).