Vehicle Localization Using Top View Image Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vehicle localization methods, especially for autonomous vehicles, face challenges in achieving high precision and reliability, as they often rely solely on GPS, which is insufficient for applications requiring decimeter-level accuracy, such as autonomous parking and driving.
Innovation Solution
A system that uses a combination of sensors like cameras, LIDAR, SAR units, and IMU to generate a top view image of the environment, which is then matched with previously captured aerial images to determine vehicle localization, incorporating image and object recognition techniques for precise positioning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS is used for vehicle localization, then the system is simple and easy to operate, but the measurement precision is insufficient for decimeter-level accuracy requirements
Solution Approach 1:
The patent combines multiple sensing systems (cameras, LIDAR, SAR units, IMU) with GPS to create a hybrid localization system. The sensed data from these multiple sources is integrated to generate top view images that are matched with aerial maps, achieving decimeter-level accuracy while maintaining system manageability through modular architecture.
Solution Approach 2:
The system employs multi-functional sensors that can perform multiple operations: cameras capture visual data for top view generation, LIDAR provides depth information, SAR units offer radar-based sensing, and IMU tracks vehicle dynamics. This multi-functionality allows a single integrated system to achieve high-precision localization through multiple complementary measurement approaches.
2Measurement precision
If multiple sensors are used to improve localization precision, then the measurement precision improves, but the device complexity increases
Solution Approach 1:
The patent segments the localization system into distinct functional modules: data acquisition module (multiple sensors), top view generation module (image processing), matching module (aerial map comparison), and output module (localization results). This segmentation allows each component to be optimized independently while maintaining overall system precision.
Solution Approach 2:
The top view image serves as an intermediary that integrates data from multiple sensors (cameras, LIDAR, SAR, IMU) and translates it into a format comparable with aerial maps. This intermediary representation simplifies the integration of heterogeneous sensor data and enables precise matching with reference maps for accurate localization.
3Reliability
If traditional localization methods are used, then the device complexity is low, but the reliability is insufficient for autonomous driving applications
Solution Approach 1:
The system generates top view images from current sensor data and compares them with previously captured aerial maps before making localization decisions. This beforehand comparison acts as a cushioning mechanism that validates the accuracy of localization estimates, ensuring reliability for autonomous driving by cross-checking against known map data before executing navigation commands.
Solution Approach 2:
The system employs feedback by continuously comparing generated top view images with stored aerial maps and using the matching results to refine and correct localization estimates. This closed-loop feedback mechanism enhances reliability by constantly validating the vehicle's position against known environmental features, ensuring accurate localization for autonomous operation.
Data Source
AI summary
A method and system of vehicle localization. Sensed data representing an environment of the vehicle is received from at least one sensor. A top view image of the environment is generated from the sensed data representing the environment of the vehicle. A vehicle localization is determined using a top view from a plurality of previously captured top view images that matches the generated top view and the sensed data. An output is generated including the vehicle localization.