When Ultrasonic Sensors and Computer Vision Join Forces for Efficient Obstacle Detection and Recognition

Area of Science:

• Assistive Technology
• Computer Vision
• Machine Learning

Background:

• Globally, 75 million people are blind and 250 million are visually impaired (VI).
• Developing Electronic Travel Aid (ETA) systems is crucial for enhancing the safe navigation and environmental awareness of VI individuals.

Purpose of the Study:

• To introduce a novel wearable assistive device for autonomous navigation of blind and VI people in dynamic urban settings.
• To improve the mobility and environmental cognition of visually impaired individuals through advanced technology.

Main Methods:

• The system integrates ultrasonic sensors and a smartphone's video camera.
• It employs computer vision and machine learning to accurately detect static and dynamic objects of any size or shape.
• Semantic interpretation of environmental data and acoustic feedback for hazard alerts are key features.

Main Results:

• Extensive objective and subjective evaluations were conducted with 21 visually impaired subjects.
• The prototype demonstrated high accuracy in identifying environmental elements and potential hazards.
• Users reported increased mobility and ease of use.

Conclusions:

• The proposed wearable device significantly aids autonomous navigation for the visually impaired.
• The system is user-friendly, easy to learn, and enhances mobility in complex environments.
• This technology offers a promising solution for improving the independence of people with visual disabilities.