V2V App (Android)

Production-oriented Android demo for real-time Vehicle-to-Vehicle (V2V) telemetry sharing and on-device trajectory prediction.

Overview

V2V App turns Android phones into cooperative vehicle nodes. Each node publishes local motion telemetry, discovers nearby peers, exchanges state at low latency, and runs local ML inference to predict short-horizon trajectory.

This is designed for demonstrations, prototyping, and field-style validation of multi-device V2V behavior.

Key Capabilities

• Real-time telemetry loop at 10 Hz (100 ms cadence)
• Peer discovery + connection using Google Nearby Connections (P2P_CLUSTER)
• Bi-directional state exchange (JSON payloads)
• On-device TensorFlow Lite inference (vehicle_trajectory.tflite)
• Premium multi-page live UI (cosmic/neon theme):
  • Dashboard tab: live radar + network telemetry + model summary
  • Devices tab: exact connected Nearby endpoint IDs
  • Coordinates tab: exact 30 predicted (x, y) coordinates

Direct APK Download

Download latest APK

Technology Stack

• Kotlin + Android SDK (minSdk 26, targetSdk 34)
• Google Play Services Nearby (play-services-nearby)
• Google Play Services Location (play-services-location)
• TensorFlow Lite Runtime (com.google.ai.edge.litert)

System Architecture

1. Telemetry Ingestion

• GPS (FusedLocationProviderClient) for position/speed
• Device orientation (TYPE_ROTATION_VECTOR) for heading
• Linear acceleration (TYPE_LINEAR_ACCELERATION) for acceleration magnitude

2. V2V Networking Layer

• Every device advertises and discovers peers simultaneously
• Connection arbitration avoids dual connection races
• Local state broadcast to connected peers at 10 Hz
• Peer tracks retained in bounded history buffers (50 states)

3. Inference Layer

• Builds ego/node/edge tensors from recent history
• Runs TFLite model inference on-device
• De-normalizes output to world coordinates for rendering

4. UI Layer

• Styled cosmic-metallic experience with animated glow accents
• Dashboard page for radar + health counters + output summary
• Devices page for exact connected endpoint IDs
• Coordinates page for exact 30 live trajectory points

Runtime Permissions

Required for full operation:

• ACCESS_FINE_LOCATION
• BLUETOOTH_SCAN (Android 12+)
• BLUETOOTH_CONNECT (Android 12+)
• BLUETOOTH_ADVERTISE (Android 12+)
• NEARBY_WIFI_DEVICES (Android 13+)

If denied, V2V discovery/exchange will not initialize.

Multi-Device Demo SOP (Standard Operating Procedure)

Use 2+ Android phones (recommended 3-5 for stronger demo impact).

1. Install the APK on all devices.
2. Enable Bluetooth and Location on all devices.
3. Keep devices within close proximity during discovery.
4. Open app on each device and grant all requested permissions.
5. Wait 5-20 seconds for discovery and connection handshake.
6. Validate live network health in status text:

• found > 0: peers discovered
• conn > 0: active links established
• tx/rx counters increasing: telemetry flowing

7. Open Devices tab to verify exact connected endpoint IDs.
8. Open Coordinates tab to monitor exact 30 predicted points.
9. Move devices and observe live radar + prediction updates.

Operational Guidance

• Keep screens awake during demonstration to reduce background throttling.
• Prefer outdoor/open spaces for better GPS stability.
• Ensure Google Play Services is available/updated on test devices.
• Use consistent app version across all participating devices.
• Swipe across tabs during demo:
  • Dashboard: operational command view
  • Devices: connectivity validation view
  • Coordinates: model precision view

Troubleshooting

• No peers found (found=0):
  • Check Bluetooth/Location enabled on all devices.
  • Re-launch app on all nodes.
  • Bring devices closer.
• No active connection (conn=0):
  • Wait for discovery handshake; watch err in status.
  • Confirm all runtime permissions are granted.
• Prediction not updating:
  • Verify telemetry is changing (move device).
  • Check rx/tx counters for data flow.

Current Scope and Limits

• Optimized for prototype/demo usage, not certified vehicular safety deployment.
• Uses mobile sensors and Nearby transport, so performance depends on device hardware and environment.

Repository Structure

• app/src/main/kotlin/com/aryan/v2v/MainActivity.kt - app orchestration loops
• app/src/main/kotlin/com/aryan/v2v/V2VManager.kt - Nearby networking + peer tracking
• app/src/main/kotlin/com/aryan/v2v/TelemetryProvider.kt - sensor + location ingestion
• app/src/main/kotlin/com/aryan/v2v/TrajectoryPredictor.kt - model input/output pipeline
• app/src/main/kotlin/com/aryan/v2v/RadarView.kt - visualization layer
• app/src/main/kotlin/com/aryan/v2v/ui/* - multi-tab premium UI fragments + adapters