Fire and Smoke Detection for One of the Busiest Dual-Bore Tunnel in a Major U.S. State

Project Background: Fire and Smoke Detection for One of the Busiest Dual-Bore Tunnel

In collaboration with a U.S.-based integration partner, Noctuai was engaged to implement a fire and smoke detection system in a heavily trafficked tunnel. The tunnel required a robust detection solution to meet safety standards and protect public safety. With the integration partner, Noctuai undertook the complex task of adapting the technology for this unique environment.

Addressing Challenges in Tunnel Fire and Smoke Detection

1. Lighting and False Alarms: The tunnel’s variable lighting, with vehicle brake lights sometimes appearing as fire sources, led to numerous false alarms in the early testing phases.
2. Environmental Sensitivity: Humidity and particulate levels from exhaust and dust created additional challenges, with the system initially misidentifying these as signs of smoke.
3. Integration with VSM through MQTT: The solution required seamless integration with the client’s Video Surveillance Management (VSM) system, using the MQTT protocol to enable real-time monitoring and data exchange.

Solution Approach: Tailored AI for Reliable Fire Detection

With the integration partner’s support, Noctuai undertook customization and real-world testing:

1. Re-Training Algorithms: The Noctuai AICam application was refined to recognize specific tunnel lighting and vehicle behavior. The team adjusted its sensitivity to avoid false alarms from car lights and ambient conditions, extensively testing under various lighting scenarios.
2. Environmental Adaptations: The smoke detection model was further optimized to differentiate between genuine smoke and common tunnel particulates, incorporating custom sensors to more precisely manage humidity and air quality.
3. Multi-Camera System with Cross-Validation: A 40-camera system was installed throughout the tunnel, with overlapping views that allowed each camera to cross-validate the others’ inferences. This setup significantly reduced false alarms, as the system relied on multiple cameras confirming the presence of smoke or fire before triggering alerts.
4. Integration through MQTT: The detection system was linked with the client’s VSM, utilizing MQTT for seamless communication and immediate incident reporting. This ensured operators received accurate, timely data on potential fire risks.
5. Rigorous Field Testing: The tunnel’s thorough acceptance testing phase allowed for continuous operational adjustments, confirming stability and reliable detection without false alarms.

Key Outcomes: Improved Detection Accuracy and Safety Standards

After a 9-month implementation and testing period, Noctuai’s fire and smoke detection solution was fully operational in the tunnel, delivering notable improvements:

• Enhanced Detection Accuracy: The cross-validation from the multi-camera system effectively reduced false positives, meeting stringent state safety standards.
• Improved Safety Measures: Reliable detection enhanced the tunnel’s safety, offering users peace of mind and aligning with regulatory compliance.
• Efficient Real-Time Monitoring: Integration with VSM enabled seamless data flow, allowing instant alerts and responsive monitoring.
• Tunnel Fire Smoke Detection application available on the AI Cam platform

This project highlights the strength of Noctuai’s collaboration with its U.S. partner, delivering a solution adapted to the most challenging environmental conditions and advancing public safety in critical infrastructure.