ANT - AI Based Firefighting Robot
Overview
A firefighting robot powered by Artificial Neural Networks (ANNs) was developed as part of a final-year research project in Artificial Intelligence. The project, managed and executed entirely by the researcher, demonstrated effective alignment with its objectives to deliver a functional prototype. The robot was designed for autonomous navigation and fire detection, utilizing an Arduino UNO microcontroller for its compatibility with sensors and hardware components. Two feed-forward neural networks were implemented, trained using gradient descent with sigmoid activation functions. The navigation neural network processes inputs from four ultrasonic sensors to determine movement directions (left, right, forward, backward) through four input nodes, five hidden nodes, and four output nodes, achieving an accuracy of 96%. The fire prediction neural network classifies the presence of fire by analyzing inputs from three types of sensors (smoke, heat, and light) through three input nodes, four hidden nodes, and two output nodes, with an accuracy of 90%. The researcher managed the entire process, including system design, neural network architecture, sensor integration, and performance evaluation, ensuring the successful realization of the project’s goals and showcasing the practical application of AI in disaster management.
My Tasks
- Built a robotic prototype capable of autonomous fire detection and response using Artificial Neural Networks (ANNs).
- Designed and trained two separate networks: one for movement control (ultrasonic sensors), and one for fire detection (heat, smoke, light sensors).
- Integrated the ANN models on Arduino UNO hardware to allow for real-world deployment and autonomous decision-making
Technologies
Arduino
C++
IoT