Publication Type : Conference Paper
Publisher : IEEE
Source : 2025 International Conference on Power, Instrumentation, Control, and Computing (PICC)
Url : https://doi.org/10.1109/picc67314.2025.11291569
Campus : Amritapuri
School : School of Engineering
Department : Electrical and Electronics
Year : 2025
Abstract : Visually impaired individuals often encounter significant c hallenges when navigating indoor e nvironments, where traditional aids such as canes provide only limited spatial awareness. This paper presents a wearable assistive system in the form of a smart backpack that delivers real-time auditory guidance through integrated sensing and on-device intelligence. Two Raspberry Pi cameras capture video from dual perspectives for object recognition using a lightweight EfficientDet L ite1 model on a Raspberry Pi 5, achieving approximately 90% precision under well-lit conditions. Four ultrasonic sensors, managed by an ESP32 microcontroller, measure proximity to obstacles and communicate distances via Bluetooth Low Energy, triggering timely audio alerts when hazards are detected within 1.5 meters. For indoor positioning, the system uses HM10 BLE beacons and RSSI-based trilateration to estimate the user's location with an average error of ±0.3 meters in a 4.5m×5m test area. All processing including object detection, obstacle avoidance, and indoor positioning and navigation, runs entirely offline. Voice feedback is delivered through a text-to-speech engine, allowing users to select navigation targets by speaking index values. Preliminary tests confirm the d evice's f easibility for slow-paced, hands-free indoor navigation, offering a practical, modular solution to enhance independence and safety for visually impaired users.
Cite this Research Publication : Navaneeth M Suresh, Archana S, Midhun M, Rekha P Nair, An Assistive Device for the Visually Impaired, 2025 International Conference on Power, Instrumentation, Control, and Computing (PICC), IEEE, 2025, https://doi.org/10.1109/picc67314.2025.11291569