D.E.B.R.I.S (Detection and Exploration Bot for Rescue, Investigation, and Survival) :

• Combines advanced sensing technologies and real-time data transmission
• Data transmitted to a web interface on a connected server
• Equipped with cameras, proximity sensors, and environmental sensors

My Role

As the AI developer, my contributions included:

• Designing and training a CNN-based thermal classification model to identify human presence from AMG8833 sensor data.
• Implementing YOLOv8 for real-time object detection using the ESP32-CAM.
• Integrating AI models into the embedded system to enable efficient processing and real-time detection.
• Collaborating with the web development team to ensure proper visualization of detection results.

---

Key Highlights:

• Advanced AI Integration: Real-time human detection using cameras and thermal imaging.
• Multi-Sensor Capabilities: Utilizes environmental sensors like DHT11, MQ2, microwave Doppler radar, and GPS for comprehensive situational awareness.
• User-Interface: Displays live video, sensor readings, and location tracking on a responsive web dashboard.
• Robust Mobility: High-torque motors and wheels for tough land to enable navigation through debris.

---

Features

• Thermal Imaging: Detects human presence in low-visibility conditions with AMG 8833 thermal camera.
• Real-Time Detection: Employs YOLOv8 for live video-based object detection and a CNN for thermal image classification.
• Environmental Monitoring: Tracks temperature, humidity, motion, and gas levels using advanced sensors.
• Web Dashboard: Provides a clear, real-time interface for monitoring data and live camera feeds.

---

AI Models Workflow

1. Thermal Image Processing:
   • Captures thermal images to make our own dataset contain humans and No humans.
   • Processes data through a CNN for human classification.
   • Customized colormap range between (25 - 35)

2. Object Detection:
   • Detects humans in real-time using YOLOv8 with live video on our server.

Web Interface

• Live Dashboard:

System Architecture

1. Embedded Systems:

   • Integration of multiple sensors including AMG8833, DHT11, MQ2 gas sensor, microwave Doppler radar, and GPS.
   • Real-time data acquisition and transmission via ESP32-CAM.

2. Artificial Intelligence:

   • Thermal Image Classification: Uses a custom CNN for detecting human presence from AMG8833 data.
   • Object Detection: Employs YOLOv8 for real-time human detection using the ESP32-CAM.

3. Web Development:

   • Displays live camera feeds, thermal images, and environmental sensor readings.
   • Provides real-time location tracking through GPS.

---

Hardware Components

• AMG8833 Thermal Camera: Detects infrared radiation for human presence detection.
• ESP32-CAM: Streams live video and integrates with YOLOv8 for real-time object detection.
• DHT11 Sensor: Measures temperature and humidity.
• MQ2 Gas Sensor: Detects combustible gases and smoke.
• Microwave Doppler Radar (RCWL-0516): Detects motion through frequency shifts.
• High-Torque Motors and Wheels: Ensures mobility over rough terrain.
• 12V Lithium-ion Battery: Provides reliable power with reverse polarity protection and filtration circuits.

---

Results

• Enhanced Rescue Efficiency: Successfully detected humans under low-visibility conditions using thermal imaging and real-time video.

• Data Visualization: Displayed sensor readings, thermal data, and location tracking on a responsive web server.

• AI Integration: Achieved high accuracy in human detection using deep learning models.

---

Acknowledgments

This project was supervised by Professor Abeer twakol and supported by resources from Mansoura University.

---

Team Members