Vehicle Localization Using Top-View Image Matching Without HD Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating high-definition digital maps for autonomous vehicles is time-consuming and resource-intensive, which hinders efficient localization of vehicles.
Innovation Solution
A vehicle computer receives a first top-view image from an external sensor and generates a second top-view image from onboard cameras, comparing them to estimate the vehicle's location without requiring HD map generation, using image processing techniques to identify the best match region and determine the vehicle's location and heading.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-definition digital maps are generated for autonomous vehicle localization, then localization accuracy is improved, but time consumption and computational resource usage increase significantly
Solution Approach 1:
The patent extracts and utilizes only the essential geometric features (road boundaries, intersections, landmarks) from the environment that are necessary for localization, rather than generating complete high-definition digital maps. This selective extraction of critical spatial information maintains localization accuracy while dramatically reducing processing time and computational resources.
Solution Approach 2:
The patent segments the localization process into independent geometric feature detection and matching stages. By dividing the complex map generation task into discrete geometric element identification (road edges, intersection points, landmark positions), the system achieves accurate localization without the computational burden of generating full HD maps.
2Measurement precision
If high-definition digital maps are generated for autonomous vehicle localization, then localization accuracy is improved, but computational resource usage increases significantly
Solution Approach 1:
The system extracts only the minimal necessary geometric features required for localization accuracy, avoiding the computational expense of generating complete high-definition digital maps. This selective approach maintains measurement precision while significantly reducing energy consumption and computational resource usage.
Solution Approach 2:
Instead of investing heavy computational resources to generate persistent high-definition digital maps, the system uses lightweight, disposable geometric feature extractions that are computed on-demand and discarded after use, dramatically reducing computational resource requirements while maintaining localization accuracy.
3Loss of information
If traditional HD map generation methods are used for vehicle localization, then comprehensive spatial information is obtained, but the process becomes time-consuming and resource-intensive
Solution Approach 1:
The patent extracts only the critical geometric spatial information (road boundaries, intersections, landmarks) necessary for accurate localization, eliminating the need to process and store comprehensive HD map data. This selective extraction maintains essential spatial information while dramatically improving localization efficiency and productivity.
Data Source
AI summary
A computer includes a processor and a memory, and the memory stores instructions executable by the processor to receive a first top-view image of an area obtained from a sensor external to a vehicle, generate a second top-view image of the area based on data from a camera of the vehicle, and determine an estimated location of the vehicle in the area based on comparing the first top-view image and the second top-view image.