SAARTHI: AI-Powered Smart Mobility Aid for Visually Challenged Individuals Using Hybrid and Priority-Based Obstacle Detection

##plugins.themes.bootstrap3.article.sidebar##

pdf
Published Apr 11, 2026

##plugins.themes.bootstrap3.article.main##

Taniya U. Katigar
Student, Department of Artificial Intelligence and Machine Learning, Dr. Bapuji Salunkhe Institute of Engineering and Technology, Kolhapur, Maharashtra, India.
Anushka S. Otari
Student, Department of Artificial Intelligence and Machine Learning, Dr. Bapuji Salunkhe Institute of Engineering and Technology, Kolhapur, Maharashtra, India.
Ankita A. Nangnure
Student, Department of Artificial Intelligence and Machine Learning, Dr. Bapuji Salunkhe Institute of Engineering and Technology, Kolhapur, Maharashtra, India.
Amruta M. Kate
Head of Department, Department of Artificial Intelligence and Machine Learning, Dr. Bapuji Salunkhe Institute of Engineering and Technology, Kolhapur, Maharashtra, India.

Abstract

Saarthi is an AI-powered smart mobility aid designed to improve safe and independent navigation for visually challenged individuals. The system combines computer vision and ultrasonic sensing within a unified architecture to detect, classify, and prioritize obstacles in real time. Unlike traditional white canes, which detect only ground-level obstacles through physical contact, Saarthi provides intelligent environmental awareness through hybrid and priority-based obstacle detection.


The system performs real-time object detection using the YOLOv8 deep learning model and captures visual input through a Raspberry Pi camera module. Multiple ultrasonic sensors are used to measure obstacle distance and improve close-range hazard detection. A sensor fusion and decision logic module combines object classification with proximity measurements to assign each obstacle a priority level such as High, Medium, or Low. Based on this prioritization, the system generates adaptive audio and vibration alerts to warn the user effectively.


The complete system operates offline on Raspberry Pi 5, ensuring low latency, privacy, and reliable real-time performance. Saarthi transforms assistive mobility from basic obstacle detection into an intelligent, context-aware navigation support system, improving safety, confidence, and independence for visually challenged users.

##plugins.themes.bootstrap3.article.details##

Issue
Vol 13 No 4 (2026): Volume 13 Issue 4
Section
Articles

References

[1]. J. Redmon, S. Divvala, R. Girshick, and A. Farhadi,
"You Only Look Once: Unified, Real-Time Object Detection," in Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR), 2016.
[2]. Ultralytics, "YOLO v8 Documentation," Available: https://docs.ultralytics.com
[3]. Raspberry Pi Foundation, "Raspberry Pi 5 Documentation," Available: https://www.raspberrypi.com/documentation/
[4]. Open CV, "Open Source Computer Vision Library,"
Available: https://opencv.org
[5].World Health Organization, "World Report on Vision," 2019.
[6]. H. Sharma, A. Kumar, "Smart Walking Stick for Visually Impaired Using Ultrasonic Sensors," International Journal of Engineering Research, 2019.
[7]. WeWALK, "Smart Cane for Visually Impaired – Product Overview," Available: https://wewalk.io
[8]. HC-SR04 Ultrasonic Sensor Datasheet, Available: https://datasheetspdf.com
[9]. PAM8403 Audio Amplifier Datasheet, Available: https://datasheetspdf.com