Hybrid Positioning via LiDAR and Visual Landmarks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing positioning systems, particularly for autonomous vehicles, face challenges in accurately determining global positions in environments where GPS signals are unreliable, such as indoor and urban canyon settings.
Innovation Solution
A method that combines image processing and LiDAR range data to identify landmarks and estimate the position of a vehicle. This involves capturing two-dimensional images, identifying landmarks, collecting range data, mapping pixel data to range data points, and using trilateration to estimate the vehicle's position.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS is used for global position estimation, then the system is simple and works in open environments, but accuracy deteriorates in indoor and urban canyon environments due to signal attenuation and multi-path interference
Solution Approach 1:
The patent combines multiple positioning techniques (GPS, visual landmark recognition, and LiDAR range data) into a hybrid system. When GPS signals are unavailable or unreliable, the system switches to using visual landmarks identified through image processing and trilateration based on LiDAR range measurements, achieving an average position error of 0.08 meters in two-dimensional space.
2Measurement precision
If visual landmark recognition is used to replace GPS, then position accuracy can be maintained in GPS-denied environments, but the difficulty of positively identifying landmarks increases system complexity
Solution Approach 1:
The patent uses LiDAR range data as an intermediary to辅助 visual landmark recognition. By mapping pixel coordinates from images to three-dimensional point cloud data, the system creates a more robust landmark identification process that combines visual information with depth information, making landmark detection more reliable in various lighting and environmental conditions.
3Measurement precision
If single-point ranging is used for position determination, then the system is simple to implement, but position accuracy is insufficient compared to multi-point ranging methods
Solution Approach 1:
The patent segments the ranging process by selecting specific representative points from the LiDAR point cloud data corresponding to identified landmarks. Instead of using all available points, the system identifies key points (such as centroid points or characteristic points) from the point cloud that map to landmark locations in the image, simplifying the data processing while maintaining high positioning accuracy through trilateration.
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
The method enhances the accuracy of position determination by leveraging visual features and range data, achieving an average position error of 0.08 meters in two-dimensional space and improving accuracy over single-point ranging methods.
Implementation Method 1
collecting range data for the landmark including one or more data points each defining a distance from the point in space to a respective point on the landmark
Data Source
AI summary
Systems, methods, and computer program products for determining the position of a first point in space. A three-dimensional point cloud is generated that comprises a plurality of data points each defining a range from the first point in space to another point in space in a region around the first point. A two-dimensional image is captured of an area including at least some of the points to which ranges have been defined. One or more objects each corresponding to a landmark are identified in the two-dimensional image, and one or more of the data points are mapped to each object. A range to each of the one or more landmarks is determined based on the ranges defined by data points mapped to the objects. A position of the first point in space is then obtained based on known positions of the landmarks using trilateration.