Navigation Key Point Localization via Neural Network Image Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle navigation systems are inefficient in providing precise instructions to drivers, as they often divert attention from the road and fail to accurately determine navigation key points such as road sign locations and turn points, limiting the driver's ability to perform actions like turns or lane changes safely and effectively.
Innovation Solution
A computer-implemented method using an artificial neural network to determine the coordinates of navigation key points by processing camera images, translating image-related coordinates into geocentric coordinates, and displaying instructions in an augmented reality format, allowing for precise navigation key point localization even in adverse conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If speech synthesis or text display is used to provide navigation instructions, then the driver receives clear directional guidance, but the driver's attention is diverted from the road
Solution Approach 1:
The patent uses an augmented reality display as an intermediary between the navigation system and the driver. Instead of using speech synthesis or text display that requires driver attention, the system projects navigation instructions directly onto the road view through the camera display, allowing the driver to receive clear directional guidance while maintaining focus on the road.
2Ease of operation
If augmented reality display is used to show navigation instructions on camera image, then the driver can see instructions in context, but the position accuracy of navigation key points must be extremely precise
Solution Approach 1:
The patent segments the navigation key point detection into two parts: a training phase where a neural network is trained on labeled images to learn key point positions, and an inference phase where the trained network automatically detects key points. This segmentation allows the system to achieve high position accuracy by learning from extensive training data rather than relying on simple geometric calculations.
Solution Approach 2:
The patent replaces traditional mechanical or geometric methods for determining navigation key point positions with an artificial neural network-based image processing system. The neural network automatically identifies road sign locations and turn points by learning from training images, substituting complex manual or rule-based position determination with an intelligent system that achieves higher accuracy and robustness.
3Device complexity
If traditional navigation systems are used, then the system structure is simple, but the navigation instructions are not precise enough for safe driving actions
Solution Approach 1:
The patent replaces traditional navigation systems that rely on pre-stored map data and geometric calculations with an image-based neural network system. The system captures real-time camera images, processes them through a trained neural network to automatically detect navigation key points, and provides precise location information. This substitution of mechanical/geometric methods with intelligent image processing achieves superior position accuracy while maintaining practical system complexity.
Data Source
AI summary
Computer-implemented method for determining coordinates of navigation key points indicative of road sign locations and/or turn points, the method comprising: collecting, as a first training data subset, one or more first images of a first camera comprised in a mobile device; obtaining, as a second training data subset, image-related coordinates of navigation key points related to the images of the first training data subset; supplying the first training data subset and the second training data subset to an artificial neural network as a training dataset; training the artificial neural network on the training dataset to predict image-related coordinates of navigation key points indicative of road sign locations and/or turn points; capturing a second image of a second camera as an input dataset, processing the input dataset by the artificial neural network to predict image-related coordinates of navigation key points indicative of road sign locations and/or turn points.