Mobility Support with Intelligent Obstacle Detection for Enhanced Safety

In recent years, assistive technology usage among the visually impaired has risen significantly worldwide. While traditional aids like guide dogs and white canes have limitations, recent innovations like RFID-based indoor navigation systems and alternative sensory solutions show promise. Nevertheless, there is a need for a user-friendly, comprehensive system to address spatial orientation challenges for the visually impaired. This research addresses the significance of developing a deep learning-based walking assistance device for visually impaired individuals to enhance their safety during mobility. The proposed system utilizes real-time ultrasonic sensors attached to a cane to detect obstacles, thus reducing collision risks. It further offers real-time recognition and analysis of diverse obstacles, providing immediate feedback to the user. A camera distinguishes obstacle types and conveys relevant information through voice assistance. The system’s efficacy was confirmed with a 90–98% object recognition rate in tests involving various obstacles. This research holds importance in providing safe mobility, promoting independence, leveraging modern technology, and fostering social inclusion for visually impaired individuals.