Vehicle Position Estimation Using GPS and Camera Landmark Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automotive navigation systems rely heavily on GPS for positioning, which can be inaccurate in areas with insufficient or unavailable GPS signals, and do not effectively utilize visual cues for both image recognition and feature extraction to enhance navigation accuracy.
Innovation Solution
A method and apparatus that integrates a GPS receiver with a CCD camera to simultaneously estimate vehicle and landmark positions using image recognition and feature extraction, employing a Kalman filter architecture to improve navigation accuracy by incorporating imagery measurements and creating a reusable database of recognizable landmarks in real-time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If GPS receiver is used for vehicle positioning, then positioning coverage is improved, but positioning accuracy deteriorates in areas with insufficient GPS signals
Solution Approach 1:
The patent combines GPS positioning with visual positioning using a camera to form an integrated positioning system. The system merges GPS coordinates with image recognition data of landmarks to achieve accurate positioning even when GPS signals are insufficient, thereby maintaining both coverage and accuracy across different environments.
Solution Approach 2:
The patent introduces landmarks as intermediary objects that connect the vehicle's position to the geographic environment. By recognizing landmarks in images and matching them with a database, the system can determine vehicle position indirectly when direct GPS positioning is unreliable, thus maintaining positioning accuracy.
2Difficulty of detecting and measuring
If visual cues are used for image recognition only, then object detection capability is improved, but navigation accuracy enhancement is lost
Solution Approach 1:
The patent makes the visual system perform multiple functions: both image recognition for object detection and feature extraction for positioning. By extracting geometric features from recognized landmarks and using them for coordinate calculation, the system simultaneously achieves object detection and navigation accuracy enhancement with the same visual cues.
Solution Approach 2:
The patent segments the visual information processing into distinct stages: object recognition, feature extraction, and position calculation. This segmentation allows the system to use different aspects of visual cues for different purposes, enabling both object detection and accurate positioning through the same camera input.
3Measurement precision
If feature extraction is performed on recognized landmarks, then positioning precision is improved, but system complexity increases
Solution Approach 1:
The patent performs preliminary actions by pre-establishing a landmark database with known coordinates and characteristics before field operation. This pre-processing reduces the complexity of real-time positioning, as the system only needs to match observed landmarks with the pre-existing database rather than performing complex calculations from scratch.
Solution Approach 2:
The patent uses a database copy of landmark information (pre-stored coordinates and features) to simplify real-time positioning. Instead of calculating absolute positions from scratch, the system copies and matches landmark characteristics from the database, reducing computational complexity while maintaining positioning precision.
Data Source
AI summary
A navigation system with a capability of estimating positions of the platform vehicle and recognized landmarks using satellite information and camera measurements is disclosed. One aspect is to represent a landmark by one or few representative features by executing image recognition and feature extraction to find representative features inside the recognized landmark. Another aspect is to re-identify the recognized landmarks in the landmark database correlating the landmark attributes obtained by image recognition and mathematical feature characteristics obtained by feature extraction with the corresponding data in the landmark database. A further aspect is to enhance navigation accuracy through the Kalman filter architecture with additional imagery measurements of landmarks whose positions are known or previously estimated. According to the aspects noted above: (1) navigation accuracy improves with augmented imagery measurements; (2) a database of recognizable landmark attributes and associated feature characteristics can be created in real-time; (3) by re-visiting the same landmarks, navigation accuracy improves and the landmark database will be re-calibrated and augmented with new landmark data.