Camera-Based Vehicle Localization Using HD Map Edge Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurate localization due to limitations in conventional maps and GPS systems, which can result in inaccurate navigation and safety issues, especially when sensors are obscured or lack sufficient data.
Innovation Solution
The use of high-definition (HD) maps combined with camera-based localization techniques that utilize imaging systems to determine the vehicle's pose without relying on depth sensors like LiDAR, by projecting 3D edgels onto image frames and optimizing the vehicle's pose based on correspondences between edge pixels and edgels to achieve precise location determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS systems are used for vehicle localization, then the system is simple and provides broad coverage, but the localization accuracy deteriorates to 3-5 meters or over 100 meters with large error conditions
Solution Approach 1:
The patent introduces HD maps as an intermediary between GPS and the vehicle localization system. The HD maps provide high-precision reference data (lane markings, road geometry) that acts as a mediator to refine GPS coordinates into accurate vehicle pose estimates, achieving 10 cm or better localization accuracy while maintaining system feasibility
Solution Approach 2:
The patent replaces the purely satellite-based GPS mechanical system with a hybrid approach using camera-based visual odometry and HD map matching. This substitution uses optical fields (camera images) and computational geometry to determine vehicle position, achieving superior accuracy without requiring complex additional hardware
2Measurement precision
If conventional maps are used for navigation, then the system is simple to implement, but the map accuracy deteriorates and cannot provide the 10 cm or less precision required for safe navigation
Solution Approach 1:
The patent applies preliminary action by pre-processing and storing high-precision road geometry, lane markings, and road feature data in HD maps before the vehicle needs them. This advance preparation allows the vehicle to perform accurate localization by matching real-time sensor data against the pre-established HD map framework, achieving 10 cm precision without real-time computational complexity
Solution Approach 2:
The patent changes the parameter of map resolution and detail from conventional low-precision maps to high-definition maps with centimeter-level accuracy. By transforming the map data structure to include precise 3D road geometry, lane boundaries, and road features, the system achieves the required 10 cm or less precision for safe autonomous navigation
3Reliability
If vehicle sensors are used to observe the environment, then real-time data is obtained, but the sensors may be obscured by corners, rolling hills, and other vehicles limiting observation capability
Solution Approach 1:
The patent uses camera images to create a visual copy of the road environment and matches it against the HD map copy. This dual-copy approach allows the system to determine vehicle position by comparing the visual scene with the pre-stored high-precision map, providing reliable localization even when direct sensor observation of road features is obscured
Solution Approach 2:
The patent transitions from 2D GPS coordinates to 6D pose estimation (3D position + 3D orientation) by utilizing camera images and HD map geometry. This dimensional enhancement provides more comprehensive vehicle state information and improves reliability by cross-validating position across multiple spatial dimensions
4Measurement precision
If camera-based localization with HD maps is implemented, then localization accuracy improves to 10 cm or less, but the computational complexity and data processing requirements increase
Solution Approach 1:
The patent segments the HD map data into manageable components (road geometry, lane markings, road features) and processes them separately during localization. This segmentation allows efficient matching of camera features to corresponding map elements, achieving 10 cm accuracy without overwhelming computational complexity
Solution Approach 2:
The patent applies partial action by using only the necessary subset of HD map data relevant to the current vehicle location and camera field of view. Rather than processing entire map datasets, the system selectively matches features within the observable area, achieving high precision while maintaining computational efficiency
Data Source
AI summary
Camera based localization performed to determine a current pose of an autonomous vehicle without the aid of depth sensors such as LiDAR. The vehicle comprises an imaging system configured to capture image frames depicting portions of the surrounding area. Based on an initial pose of the vehicle, edgels corresponding to three-dimensional locations are loaded and mapped to corresponding edge pixels of the captured image frame. A pose of the vehicle is optimized based upon the determined correspondences by identifying a transformation that minimizes a distance between the edgels and their corresponding edge pixels. The determined transformation can be applied to the initial pose to determine an updated pose of the vehicle.