Turning Uncertainty into Predictability.
BFlow is an end-to-end infrastructure solution that transforms physical environments into a live stream of data, anticipating bottlenecks before they impact the passenger experience.
The greatest challenge in managing crowded public spaces is the "invisibility" of its dynamics. BFlow eliminates reactive management by providing a window into the future—monitoring not just current congestion, but predicting flow patterns using IoT sensors and Machine Learning.
- Non-Invasive Tracking: Anonymous Bluetooth detection (camera-free).
- Predictive Brain: Real-time waiting time estimation.
- Contextual Intelligence: OCR-driven boarding pass digitalization for flight-specific flow analysis.
- Ultra-Low Power: Built on ESP32 nodes using the ESP-NOW protocol.
- Privacy by Design: 100% GDPR compliant through hardware ID anonymization.
- Hyper-Local Engagement: Dynamic terminal maps and personalized duty-free recommendations based on real-time location.
- Microcontrollers: ESP32 nodes.
- Networking: ESP-NOW (Private mesh network).
- Signal Processing: Kalman Filter implementation for signal smoothing and proximity accuracy.
- Boarding Pass Processing: YOLOv8nano + EasyOCR for instant data extraction.
-
Predictive Modeling: Random Forest Regressor.
-
Performance:
$R^2 = 0.97$ -
Accuracy:
$MAE = 3.6$ min
-
Performance:
- Android Integration: Google Geofencing API for automated Bluetooth activation.
- Data Pipeline: Real-time correlation between sensor density and flight capacity/urgency.
- Sensing: ESP32 nodes detect anonymous Bluetooth signals and estimate proximity.
- Contextualizing: Boarding passes are scanned to link real-time density with flight specifics (capacity, gate urgency).
- Predicting: The ML model correlates sensor data with ticket context to forecast bottlenecks.
- Acting: Real-time alerts are issued for proactive staff deployment and passenger notifications.
Privacy is our core priority. BFlow adheres to Privacy by Design principles:
- No PII: We do not store facial data, names, or personal identities.
- Geofencing: Tracking is strictly limited to the terminal perimeter.
- Anonymization: All processing is performed on hashed hardware IDs.
- Compliance: Full GDPR alignment from the hardware layer up.
| Member | Role |
|---|---|
| Balahura Vlad | Frontend Dev & UI/UX Designer |
| Hordoan Darius | Full-Stack Engineer & System Integration |
| Halit Silviu | Bluetooth & Hardware Engineer |
| Moga Antonia | ML Engineer & Product Strategy |
| Stelea Sonia | Backend Dev & Hardware Engineer |
| Oniceag Diana | Data & API Engineer |