Vision-Based Mobility and Navigation Aid for the Visually Challenged People

Paper Link

The project aims to assist visually impaired individuals with navigation by providing real-time obstacle avoidance and route guidance. The system takes a voice command to set the user's desired destination, then identifies and guides the user along a complete route, notifying them when a turn is needed. Simultaneously, it executes an obstacle avoidance model that alerts the user when an obstacle is nearby. The system is designed to operate in real-time, with controlled delays between voice commands to prevent overwhelming the user, resulting in an improved navigation and obstacle avoidance experience.

---

Introduction:

• An estimated 285 million Visually Impaired people in the world
• The risk of mortality associated with Visual Impairment in a study was around 25% casualties
• Secure and Independent Mobility is a challenge for the Visually Impaired
• Mobility and Navigation Difficulties can be solved with Electronic Travel Assistance

Research Gap:

• Lack of Useful Functions
• Inaccurate Distance Estimation
• Bulky and Heavy-Weight
• Ergonomic and Aesthetic
• Steep Learning Curve
• Expensive

Proposed Solution:

• Modular Assistive System
• Consists of two Independent Modules:
  • Object Detection Module
  • Voice Interactive Orientation and Navigation Module
• Object Detection Module for Detection and Avoidance
• Navigation Module for Navigation and Path Planning
• Simplified Audio Feedback

Novelty:

• Obstacle Detection, Depth Estimation and Route Estimation
• Rerouting Capability
• Convenient Carry and Lightweight
• No Geographical Bounds
• Low Power Consumption
• Inexpensive

Methodolgy:

Architecture of the Proposed System:

Object Detection Module:

HSV Thresholding:

Navigation Module:

Results:

Study Design

• Experiments: 10
• Environment: Footpath (4m-5m Wide)
• Condition: Medium Crowded
• Subjects: Individuals with camera strapped on their chest

Combined Precision and Recall for the Detected Obstacles:

• Recall: 91.7%
• Precision: 81.5%

Advantages & Limitations:

Advantages:

• Simplified Audio Feedback
• Real-Time Implementation
• Blend of Obstacle and Navigation
• Low Power Embedded System
• Wearable System
• High Precision
• Cost Effective

Limitations:

• Required Uniform Illumination
• Low GPS Accuracy

Conclusion:

• Implemented a new method for the Visual-Based Navigation System for the Visually Impaired.
• Reduced Information Load for Mobility and Navigation Assistance
• Simplified representation of the Surrounding Environment
• Accuracy:
  • Object Detection: 91.7%
  • Orientation and Navigation: 77%

---