Wearable Vision Aid With On-Device Indoor Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing assistive technologies for people with visual impairments are limited in their functionality, often focusing on single tasks like navigation or object recognition, and struggle to provide comprehensive assistance in indoor environments without requiring server processing.
Innovation Solution
A wearable assistive device equipped with a camera, laser range finder, and single-board computer that performs real-time image processing, including scene recognition, object detection, obstacle avoidance, and text recognition, using lightweight and efficient machine learning models, all executed locally without the need for an internet connection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If server processing is used for image analysis, then recognition accuracy is improved, but system complexity and dependency on external infrastructure increase
Solution Approach 1:
The single-board computer performs image processing and object recognition locally without requiring external server processing. The system serves itself by executing machine learning models directly on the embedded device, eliminating dependency on external infrastructure while maintaining recognition capabilities
Solution Approach 2:
The system segments the processing workload by using a lightweight single-board computer for local image analysis rather than relying on a centralized server. This division allows the device to handle recognition tasks independently while reducing system complexity
2Adaptability or versatility
If comprehensive assistive functions are provided, then user assistance quality is improved, but device functionality complexity increases
Solution Approach 1:
The single-board computer serves multiple functions including image capture, processing, object recognition, text recognition, and navigation assistance. This multi-functional approach provides comprehensive assistive capabilities while consolidating complexity into a single integrated device rather than multiple separate systems
Solution Approach 2:
The system merges camera, laser range finder, and processing units into a single integrated wearable device. This consolidation provides comprehensive navigation and object recognition functions while simplifying the overall system architecture by combining multiple functionalities into one unit
3Speed
If real-time processing is implemented, then response speed is improved, but computational resource requirements increase
Solution Approach 1:
The system uses lightweight machine learning models with optimized parameters that enable real-time processing on a resource-constrained single-board computer. By changing the parameters of the neural networks to be more efficient, the system achieves real-time object recognition without requiring excessive computational resources
Solution Approach 2:
The system processes images in real-time by analyzing only the most critical features and objects in the scene rather than performing exhaustive analysis. This partial processing approach maintains real-time response speed while reducing the computational burden on the single-board computer
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables users with visual impairments to navigate and interact with their indoor environment effectively, recognizing objects and reading text in real-time, ensuring safety and independence with a lightweight, self-contained system.
Implementation Method 1
capturing an image of a real-time scene of an environment
Implementation Method 2
laser range finder
Implementation Method 3
laser range finder
Data Source
AI summary
Systems, methods, apparatuses, and computer program products for assisting people with visual impairments. A method for operating a mobile assistive device may be provided. The method may include capturing an image of a real-time scene of an environment. The method may also include sending the image to a single-board computer. The method may further include processing the image. In addition, the method may include providing navigation assistance to a user based on the processed image.