An Assistant System For Blind To Avoid Obstacles Using Artificial Intelligence Techniques

Abstract

This study focuses on developing an assistive system for blind individuals for collision avoidance of obstacles by combining artificial intelligence techniques; Convolutional Neural Networks (CNN), fuzzy logic control (FLC), and genetic algorithms(GA), This integrated system, named the (NFG) Neural Fuzzy Genetic). The proposed system combines artificial intelligence techniques through detecting and tracking objects, measuring the distance between objects and the blind person, and providing movement guidance using three ultrasonic sensors with FLC and optimization GA. The integration of these technologies offers an innovative solution to enhance the mobility and safety of blind individuals.
Specifically, object detection and tracking are applied through CNN, with an obstacle detection range of up to 40 meters. The obstacle recognition system is trained on the ResNet50 model, which includes 50 million trained images and more than 1,000 obstacle classifications, resulting in high accuracy in identifying and detecting obstacles. When tested, the accuracy of the trainer model reached 99.9%. FLC is then used to provide motor guidance and help make appropriate decisions in the presence of obstacles, navigate safely and independently, and determine movement directions in obstacle-free paths with the help of three sensors.