Degree

Bachelor of Science (Computer Science)

Department

Department of Computer Science

School

School of Mathematics and Computer Science (SMCS)

Advisor

Dr. Imran Khan, Assistant Professor, Institute of Business Administration, Karachi

Co-Advisor

Shalin Amir Ali, Folio3

Keywords

Custom Heuristics, Computer Vision, Mobile Application, SDK Integration, Assistive Technology

Abstract

This report presents the design and development of a smartphone-based navigation system aimed at assisting blind users in independently traversing the IBA University campus. The system integrates two core modules: (1) a Navigation Module that provides turn-by-turn guidance using OpenStreetMap (OSM) data and GPS positioning, and (2) an Obstacle Detection Module powered by a lightweight YOLOv8n model running on-device via TensorFlow Lite to detect hazards such as pedestrians, vehicles, and bicycles. Real-time audio feedback is provided through Google Text-to-Speech, enabling users to receive navigation instructions and obstacle alerts without relying on external hardware. The Android application was evaluated through simulated campus scenarios, focusing on usability, alert accuracy, and responsiveness. Results from these simulations indicate that the system can effectively guide users and provide timely obstacle warnings in a typical campus environment. Future improvements include real-world testing with blind individuals, enhanced obstacle detection, and the integration of indoor navigation support.

Tools and Technologies Used

Platforms & Tools:

  • Android Studio
  • OsmAnd & OpenStreetMap (OSM) SDK

Programming Language:

  • Kotlin

Libraries & Models:

  • YOLOv8n (for real-time object and obstacle detection)
  • Google Translate (for voice navigation cues)

Methodology

The development approach follows a use-case driven methodology, where each user flow is identified, built, and tested in iterative cycles. Group members divided responsibilities based on modules and use cases.

This allowed for:

  • Targeted optimization of each feature
  • Manual testing in real-world scenarios
  • Refinement through iterative feedback and improvements

Special emphasis was placed on lightweight computation, avoiding processing-heavy models and implementing custom heuristics for depth estimation to ensure the app works effectively on standard Android devices.

Document Type

Restricted Access

Submission Type

BSCS Final Year Project

Share

COinS