RTAIAED: A Real-Time Ambulance in an Emergency Detector with a Pyramidal Part-Based Model Composed of MFCCs and...

Area of Science:

• Computer Science
• Artificial Intelligence
• Traffic Engineering

Background:

• Ambulance response times are critical in emergencies.
• Traffic light systems can impede ambulance navigation.
• Existing solutions lack real-time, multi-modal detection capabilities.

Purpose of the Study:

• To introduce a novel automated Real-Time Ambulance in an Emergency Detector (RTAIAED).
• To enable timely and safe green light transitions for approaching ambulances.
• To improve overall emergency medical response efficiency.

Main Methods:

• Simultaneous processing of video and audio data.
• Video analysis using a custom YOLOv8 model inspired by Part-Based Model theory.
• Audio analysis employing a neural network for Mel Frequency Cepstral Coefficients (MFCCs).
• Logic-based integration of sensory inputs for precise AIAE identification.

Main Results:

• The RTAIAED demonstrated robust real-time performance (11.8 fps, 0.25s response time on Jetson Nano).
• Effective detection in challenging conditions (nighttime, adverse weather, noisy environments).
• Successful management of traffic signals on one-way roads and temporary situations like roadworks.

Conclusions:

• The RTAIAED system accurately detects ambulances in real-time using a multi-modal approach.
• The system enhances traffic light control for emergency vehicles, reducing response times.
• This technology offers a reliable solution for improving emergency navigation through urban traffic.