AI-Powered Smart Wearable Glass for the Visually Impaired with Voice-based Navigation and Scene Awareness

Abstract

Visually impaired people often face difficulties in navigation and understanding their surroundings independently. Traditional tools like white canes and guide dogs lack in providing information. With recent advancements in artificial intelligence and wearable technology, more effective solutions are becoming possible. In this project, we introduce a pair of AI-powered smart glasses intended to assist visually impaired individuals that deliver real-time audio descriptions of the environment. The system integrates an ESP32-S3 microcontroller with an OV2640 camera to capture images. These images are sent to a server where AI models (like gpt-4o for image captioning and gtts for speech) analyze the scene and turn it into spoken descriptions. The audio is then delivered wirelessly through a Bluetooth earpiece. Inspired by previous work such as Vis Buddy and other smart glasses, our design focuses on being low-cost, easy to use, and practical in real life. The glasses respond to button presses, making them interactive and user-friendly. User testing shows that this system helps users become more independent and aware of their surroundings. Overall, this work aims to make daily life easier and more accessible for visually impaired people using modern, affordable technology.

Introduction

The text discusses the challenges faced by visually impaired individuals in performing daily tasks such as mobility, object recognition, and navigation. Traditional aids like white canes and guide dogs provide basic support but have limitations, including restricted range, high maintenance, and lack of detailed environmental information. To address these gaps, advances in wearable technology, AI, and computer vision have enabled the development of smart assistive devices, including AI-powered smart glasses.

The proposed smart glasses system is low-cost, compact, and user-friendly, triggered by a single push-button. A camera captures the environment, an embedded processor or cloud-based AI identifies objects and scenes, and the information is converted into real-time audio feedback, enabling greater independence for users. Unlike complex multi-sensor or haptic systems, this design prioritizes minimal hardware, affordability, and core functionality, making it suitable for low-resource settings.

The literature review highlights prior innovations in wearable AI devices, object recognition, OCR, and multimodal feedback, noting trends in edge–cloud processing, low-light performance, and contextual AI. Key gaps include privacy-preserving on-device inference, standardized real-world testing, low-cognitive-load UX, and fail-safe navigation.

The design methodology includes:

• Hardware: ESP32-S3 microcontroller, OV2640 camera, Bluetooth earpiece, push button, battery.

• Software: AI models for image recognition, TTS engines (gTTS), Python/Arduino programming.

• Workflow: Image capture → AI processing (scene/object recognition) → Text-to-speech conversion → Audio feedback to user.

• Optimization: Reduce latency, improve comfort, ensure reliable operation, and refine user experience.

The system enhances mobility, situational awareness, and independence for visually impaired users while remaining cost-effective, scalable, and practical for real-world deployment.

Conclusion

The project aims to bridge the disconnect between visual impairment and independent mobility by the use of modern technologies like Artificial Intelligence, sensors, and voice interaction. Through the integration of obstacle detection, scene recognition, and voice guidance, the Smart Wearable Glass allows visually impaired users to perceive their environment in real-time. Our prototype was able to demonstrate successfully that visually impaired individuals can receive environmental feedback in the form of sound, which tells them about obstacles, objects, or locations around them. The system is an artificial visual guide, giving users verbal information and notifications.

References

[1] Evangelos Geoggarous, Diongsysia Danai Brilli, Konstantina Nikita, Nikos Melanitis, Stefania Tsilivaki, “AIris: An AI-powered Wearable Assistive Device for the Visually Impaired.” August 2024 [2] Smit Patel, Nirav Satani, Sandip Patel, “AI-Powered Glasses for Visually Impaired Persons.” International Journal of Recent Technology and Engineering (IJRTE), July 2020. [3] Mukhriddin Mukhiddinov, Jinsoo Cho, “Smart Glass System Using Deep Learning for the Blind and Visually Impaired”, November 2021. [4] Shiloah Elizabeth D, Ishwarya Sivakumar, Nishaali Meenakshisundaram, Ishwarya Ramesh, Sunil Retmin Raj, “VisBuddy- A smart wearable assistant for the visually challenged.”, Jan 2022. [5] Miguel Altamirano Cabrera, Yara Mahmoud, Luis Moreno, Muhammad Haris Khan, Dzmitry Tsetserukou, Issatay Tokmurziyev, “LLM-Glasses: GenAI-driven Glasses with Haptic Feedback for Navigation of Visually Impaired People,” August 2025. [6] Syeda Anshrah Gillani, Mahmoud Aljawarneh, Abdul Akbar Khand, Mirza Samad Ahmed Baig, Muhammad Hamzah Siddiqui, Shahid Munir Shaha, “AI-based Wearable Vision Assistance System for the Visually Impaired: Integrating Real-Time Object Recognition and Contextual Understanding Using Large Vision-Language Models”, 2023. [7] Liumei Long, Xiaochen Zhang, Xiaoyu Huang, Wujing Li, XingXu, Yiran Ding, “Advancements in Smart Wearable Mobility Aids for Visual Impairments: A Bibliometric Narrative Review,” 2024. [8] Frode Eika Sandnes, Bineeth Kuriakose, Raju Shrestha, “SceneRecog: A Deep Learning Scene Recognition Model for Assisting Blind and Visually Impaired Navigate using Smartphones” IEEE International Conference on Systems, Man, and Cybernetics, 2021. [9] V. Vijaya Baskar, Ikben Ghosh, S.Karthikeyan, R. J.Hemalatha, T.R.Thamizhvani, “An Indoor Obstacle Detector to Assist the Visually Impaired Person in Real-Time with a Navigator”, 2021 International Conference on Computational Performance Evaluation (ComPE), North- Eastern Hill University, Shillong, Meghalaya, India. Dec 1- 3, 2021. [10] Meishan Liu, Liye Zhu, Jinming Wang, Yizhou Qian, ”The Impact of Data-Driven Feedback on Programming Learning: A Cognitive Load Perspective”, 2025, the 7th International Conference on Computer Science and Technologies in Education. [11] Lope Ben Porquis, Sara Finocchietti, Giorgio Zini, Giulia Cappagli, Monica Gori, Gabriel Baud-Bovy, “ABBI: A Wearable Device for Improving Spatial Cognition in Visually-Impaired Children”, IEEE BIOCAS 2017. [12] Amrita Thapa, Sutapa Debbarma, Bijoy Kumar Upadhyaya, “Design of a Raspberry Pi-Based Electronic Travel Assistant for Visually Impaired Persons”, 2022 IEEE Calcutta Conference. [13] Esra Ali Hassan, “Smart Glasses for the Visually Impaired People”, January 2016. [14] Anurag Patil, Abhishek Borse, Anish Borse, Ajay Phad, Dr. Simran Khiani, “A Low-Cost IoT-based Navigation Assistance for Visually Impaired Persons”, Proceedings of the 8th International Conference on Communication and Electronics Systems (ICCES 2023). [15] Makanyadevi K, Avinash M V, Chandhrakiran S V, Kavitha S, Bhalaram Krishna S A, Dhiyanesh S. “AI- Powered Smart Glasses and Shoes for the Visually Impaired”. Proceedings of the 6th International Conference on Mobile Computing and Sustainable Informatics (ICMCSI- 2025). [16] Md. Repon Islam, Muhammad Sheikh Sadi, “Generative AI-assisted Standalone Wearable Reading Glasses for the Visually Impaired”, 2024 IEEE 3rd International Conference on Robotics, Automation, Artificial-Intelligence and Internet-of-Things (RAAICON). [17] Vijaya Vardan Reddy S, B.Sathyasri, J.Joselin Jeya Sheela, Malarvizhi. C Vanaja, Shyam S, “Smart Blind Assistant using Deep Learning for the Visually Impaired Users”, Second International Conference on Augmented Intelligence and Sustainable Systems (ICAISS 2023). [18] Sudarshan S, Mohana, Ambika G, Kavitha A, Nataraj K, Sudhangowda B S, “Object Detection using Vision Transformer and Deep Learning for Computer Vision Applications”, Proceedings of the 7th International Conference on Intelligent Sustainable Systems (ICISS-2025). [19] Sudhakar J, Viswesh Iyer V, Sree Sharmila T, “Image Caption Generation using Deep Neural Networks”, 2022 International Conference for Advancement in Technology (ICONAT). [20] Sindhwani, Srikanthan, “Framework for Automated Application-Specific Optimization of Embedded Real-Time Operating Systems”, International Conference(ICICS 2005). [21] R. K. Yadav, Anurag Singh, “Comparative Study of PEGASIS-based Protocols in Wireless Sensor Networks”, International Conference,2016. [22] Zainab Rasheed, Sameh Ghwanmeh, Aziz Almahadin, “A CRITICAL REVIEW OF AI-DRIVEN AR GLASSES FOR REAL TIME MULTILINGUAL COMMUNICATION”, 2024 Advances in Science and Engineering Technology International Conferences (ASET). [23] Zhihua Lu, Philipp Heidenreich, Abdelhak M. Zoubir, “OBJECTIVE QUALITY ASSESMENT OF SPEECH ENHANCEMENT ALGORITHMS USING BOOTSTRAP- BASED MULTIPLE HYPOTHESIS TESTS.”, International Conference (ICASSP 2010). [24] Sai Dikkatwar, Abhishek Mhaske, Harsh Ninawe, Harshit Khadia, Shraddha S.Gugulothu, “Recent Trends in Low Light Object Detection in Indoor Spaces for Assisting the Visually Impaired”, 2025 3rd International Conference on Intelligent Data Communication Technologies and Internet of Things (IDCIoT). [25] Weijun Wang1, Junquan Wang, Zhiqiang Zhang, Da Shi, “OpenCV Implementation of Image Processing Optimization Architecture of Deep Learning Algorithm based on Big Data Processing Technology”, Proceedings of the International Conference on Sustainable Computing and Data Communication Systems (ICSCDS-2022). [26] Poorva Agrawal, Kashish Sharma, Keyur Dhage, Isha Sharma, Nitin Rakesh, Gagandeep Kaur, “Speech-to-Text Conversion and Text Summarization”, 2024 First International Conference on Technological Innovations and Advanced Computing (TIACOMP).

Copyright

Copyright © 2025 Dr. Gauri Ghule, Manasvi N. Pawar, Yash G. Patil, Pushpraj R. Patil, Dhanashri G. Pawar. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.