Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ForestProtector: An IoT Architecture Integrating Machine Vision and Deep Reinforcement Learning for Efficient Wildfire Monitoring

Kenneth Bonilla-Ormachea, Horacio Cuizaga, Edwin Salcedo, Sebastian Castro, Sergio Fernandez-Testa, Misael Mamani

TL;DR

ForestProtector tackles the need for scalable, affordable wildfire surveillance over large forest areas by integrating IoT sensor networks with a central gateway that uses deep reinforcement learning to steer a 360° camera and a 3D-CNN to verify smoke observations. The approach prioritizes high-risk sectors using multi-sensor signals and a learned policy, while keeping costs low through LPWAN-enabled IoT nodes and edge processing. Key contributions include the DRL-driven sector selection mechanism, a dataset and five 3DCNN architectures for smoke detection, and a cloud-backed alert and visualization pipeline. Field experiments in Bolivia demonstrate timely alerts and underline trade-offs between DRL responsiveness and CNN inference time, pointing to practical potential and avenues for optimization in resource-constrained settings.

Abstract

Early detection of forest fires is crucial to minimizing the environmental and socioeconomic damage they cause. Indeed, a fire's duration directly correlates with the difficulty and cost of extinguishing it. For instance, a fire burning for 1 minute might require 1 liter of water to extinguish, while a 2-minute fire could demand 100 liters, and a 10-minute fire might necessitate 1,000 liters. On the other hand, existing fire detection systems based on novel technologies (e.g., remote sensing, PTZ cameras, UAVs) are often expensive and require human intervention, making continuous monitoring of large areas impractical. To address this challenge, this work proposes a low-cost forest fire detection system that utilizes a central gateway device with computer vision capabilities to monitor a 360° field of view for smoke at long distances. A deep reinforcement learning agent enhances surveillance by dynamically controlling the camera's orientation, leveraging real-time sensor data (smoke levels, ambient temperature, and humidity) from distributed IoT devices. This approach enables automated wildfire monitoring across expansive areas while reducing false positives.

ForestProtector: An IoT Architecture Integrating Machine Vision and Deep Reinforcement Learning for Efficient Wildfire Monitoring

TL;DR

ForestProtector tackles the need for scalable, affordable wildfire surveillance over large forest areas by integrating IoT sensor networks with a central gateway that uses deep reinforcement learning to steer a 360° camera and a 3D-CNN to verify smoke observations. The approach prioritizes high-risk sectors using multi-sensor signals and a learned policy, while keeping costs low through LPWAN-enabled IoT nodes and edge processing. Key contributions include the DRL-driven sector selection mechanism, a dataset and five 3DCNN architectures for smoke detection, and a cloud-backed alert and visualization pipeline. Field experiments in Bolivia demonstrate timely alerts and underline trade-offs between DRL responsiveness and CNN inference time, pointing to practical potential and avenues for optimization in resource-constrained settings.

Abstract

Early detection of forest fires is crucial to minimizing the environmental and socioeconomic damage they cause. Indeed, a fire's duration directly correlates with the difficulty and cost of extinguishing it. For instance, a fire burning for 1 minute might require 1 liter of water to extinguish, while a 2-minute fire could demand 100 liters, and a 10-minute fire might necessitate 1,000 liters. On the other hand, existing fire detection systems based on novel technologies (e.g., remote sensing, PTZ cameras, UAVs) are often expensive and require human intervention, making continuous monitoring of large areas impractical. To address this challenge, this work proposes a low-cost forest fire detection system that utilizes a central gateway device with computer vision capabilities to monitor a 360° field of view for smoke at long distances. A deep reinforcement learning agent enhances surveillance by dynamically controlling the camera's orientation, leveraging real-time sensor data (smoke levels, ambient temperature, and humidity) from distributed IoT devices. This approach enables automated wildfire monitoring across expansive areas while reducing false positives.
Paper Structure (16 sections, 5 equations, 7 figures, 4 tables)

This paper contains 16 sections, 5 equations, 7 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Proposed System
  4. IoT Sensor Nodes
  5. Central Gateway and Cloud Server
  6. Deep Reinforcement Learning Model
  7. Model design and training
  8. Decision-making process for detecting wildfires
  9. Wildfire Detection Using Computer Vision
  10. Dataset preparation
  11. 3DCNN Architecture
  12. Experimental Results and Discussion
  13. Deep Reinforcement Learning Model
  14. Smoke Detection
  15. System Performance Results
  16. ...and 1 more sections

Figures (7)

  • Figure 1: Overview of the ForestProtector architecture for efficient wildfire monitoring.
  • Figure 2: System's software architecture.
  • Figure 3: S5, the best performing 3DCNN model for smoke detection at long distances.
  • Figure 4: Moving Average of Rewards and Training Loss over Time, showing the agent's learning progress and convergence.
  • Figure 5: Performance metrics of the S5 model, including loss, accuracy, AUC, and confusion matrix.
  • ...and 2 more figures