Assisted Pedestrian Crossing System for Visually Impaired People Through Machine Vision Recognition and LoRa Communication

This article presents the design and implementation of a pedestrian crossing system focused on people with visual impairments. The proposed solution combines artificial vision using a customized YOLOv5 model with wireless communication between the system and a smart cane based on LoRa technology. It begins with a crossing request when the user approaches a traffic light equipped with the system within a set range. Upon receiving this request, an ESP32-CAM module captures an image and communicates with a PC running the YOLOv5 program to detect the presence of people with the following characteristics: a person with a cane that integrates a LoRa-compatible device and dark glasses. Communication between the ESP32 microcontroller and the PC is carried out via the UART port, ensuring low-latency interaction. The system offers a cost-effective, scalable, and inclusive solution for smart city environments, improving the mobility and autonomy of people with visual impairments. After implementation, a 60% reduction in incidents at high-risk intersections and urban accessibility is estimated.