omni track

The ultimate B2B fleet orchestration console integrating real-time multi-provider telematics, AI cabin vision, and precision fuel intelligence.

Comment

Inspiration

In the modern logistics and transportation sectors, fleet visibility is paramount. However, fleet operators face three massive, disconnected challenges:

  1. Fragmented Telematics: Mixed fleets run on heterogeneous tracking hardware (e.g., Protrack, IOPGPS), forcing operators to manage multiple tracking portals, distinct credentials, and inconsistent API schemas.
  2. Disconnected Cabin Video Monitoring: Cabin dashcams and driver-safety monitoring cameras exist in isolated silos, preventing dispatchers from correlating critical driving telemetry (like geofence breaches or harsh braking) with visual evidence in real time.
  3. Lack of Fuel Visibility: Fuel represents the largest operational expense. Without real-time fuel telemetry, detecting fuel siphoning (theft) or optimizing refueling cycles is impossible.

Inspired by the H0: Hack the Zero Stack challenge, we realized we no longer had to choose between shipping a beautiful, interactive product quickly and utilizing data infrastructure designed for real-world production scale. By leveraging Vercel v0 to scaffold a premium Next.js dashboard and connecting it to Amazon Aurora PostgreSQL, we unified telematics normalization, live AI cabin video streaming, and precision fuel intelligence into one cohesive, production-ready B2B platform.

What it does

OMNI is a unified, real-time B2B fleet orchestration console that abstracts away hardware-specific constraints and consolidates operations into a single pane of glass:

  • Unified Telematics Sync: Automatically polls and aggregates vehicle positions and device status from multiple telematics providers (Protrack and IOPGPS) every 5 seconds.
  • Tactical Live Tracking Map: An interactive map showing vehicle paths, current speed, ignition states, and active provider filters.
  • AI Cabin Vision: Routes live video feeds from driver-monitoring cameras, showing cabin status and generating event-based distraction alerts.
  • Precision Fuel Visibility: Real-time fuel gauge tracking with integrated consumption calculations and automated fuel siphoning (theft) warnings.
  • Safety-Locked Engine Cutoff: Allows dispatchers to remotely arm or disarm engine immobilizer relays, protected by a safety gateway that locks the cutoff command if the vehicle is in motion.

How I built it

The project leverages a modern, highly decoupled architecture built for rapid deployment and massive scale:

  1. Frontend Scaffolding with Vercel v0: We used Vercel's v0 to rapidly scaffold and design the Next.js (React 19) B2B console. This allowed us to build an aesthetic, premium dark/light mode tactical map interface, live camera players, and interactive fuel telemetry dashboards in record time.
  2. Django Ninja Backend API: The backend service runs on Python 3.11 with Django Ninja, providing type-safe schemas and high-speed JSON serialization.
  3. Flutter Mobile Client: A high-fidelity mobile app built for ground operators with real-time mapping, circular geofence configurations, and safety engine controls.
  4. Data Foundation - Amazon Aurora DB: For database scaling, we provisioned an Amazon Aurora PostgreSQL cluster (deployed in the eu-north-1 Stockholm region, configured with a primary writer and a reader instance for high availability). Aurora easily handles high-throughput writes from real-time GPS polling and continuous fuel sensor updates.
  5. Real-time Video & Sensor Ingestion: Integrated WebRTC and HLS streaming protocols for low-latency cabin camera feeds, alongside normalized CAN-bus / capacitive sensor inputs for fuel reporting.

Challenges I ran into

  • Normalizing Inconsistent API Handshakes: Telematics providers implement unique authentication and timestamping mechanisms. We had to build a custom TelematicsProviderClient interface to handle different signature hashes, such as: $$\text{Signature}_{\text{Protrack}} = \text{MD5}(\text{MD5}(\text{password}) + \text{timestamp})$$
  • Live Video Bandwidth Constraints: Streaming live video from cabin monitoring cameras over cellular networks in remote areas is highly bandwidth-constrained. We implemented dynamic resolution scaling and event-based recording thresholds, triggering HD uploads only when the AI camera's local distraction detection confidence exceeds the safety threshold: $$\text{Confidence}{\text{event}} \ge \theta{\text{threshold}}$$
  • Safety Lock Engineering: Allowing remote engine immobilization is critical, but executing it on a moving vehicle could cause accidents. We built a speed-locked Safety Gateway to evaluate the velocity constraint before dispatching a cutoff instruction: $$v_{\text{current}} < v_{\text{threshold}} \quad (v_{\text{threshold}} = 5\text{ km/h})$$

Accomplishments that I'm proud of

  • The Safety Gateway Speed Lock: Creating a reliable, hardware-integrated safety override that prevents dangerous engine shutoffs when a vehicle is moving $\ge 5\text{ km/h}$.
  • Zero-Lag Telemetry Normalization: Normalizing heterogeneous streams from multiple third-party systems and storing them instantly without database write lock bottlenecks.
  • Premium UX/UI Integration: Designing a gorgeous, responsive dark/light mode interface using Vercel v0 that looks and feels like a modern military tactical terminal.
  • Resilient Offline Failover: Building a high-fidelity offline simulation system into both the mobile app and backend mock telemetry generators to ensure absolute testability even when external provider APIs are down.

What I learned

  • The Zero Stack Velocity: Vercel v0 proved that beautiful, highly functional frontends can be scaffolded in minutes, allowing developers to focus their efforts on data design and backend integration.
  • Serverless Database Power: Deploying Amazon Aurora in a multi-AZ cluster configuration showed us how easily Postgres handles high-throughput write streams from heterogeneous tracking hardware.
  • Decoupling Data Layer: Decoupling the telematics APIs from our core application logic via a normalized abstraction layer is key to managing highly fragmented IoT environments.

What's next for omni

  • Universal Telematics Adapter: Adding native integration for other major telematics hardware protocols (Concox, Coban, Teltonika) to make OMNI the ultimate universal fleet gateway.
  • Predictive Fuel Analytics: Integrating machine learning algorithms (via Amazon SageMaker) to predict fuel theft patterns and optimize refueling stops based on historical route telemetry.
  • Automated Emergency Dispatch: Configuring automated dispatcher alerts and incident logging triggered directly by AI camera safety events (e.g. driver drowsiness detection or severe collision events).

Built With

Share this project:

Updates