Lane-Level Vehicle Localization Using Dead Reckoning and Image Sensing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Global Navigation Satellite Systems (GNSS) such as GPS can be inaccurate or unavailable in urban areas due to signal blocking, making it difficult for autonomous and semi-autonomous vehicles to determine precise lateral position within a lane, which is crucial for advanced driver-assistance systems like lane keeping assistance and lane departure warning.
Innovation Solution
A vehicle navigation system using standard-definition map data, image sensors, and a processor to estimate a vehicle's initial location via dead reckoning, detect lateral position within a lane using image data, and update the location based on the difference between estimated and actual positions, enabling lane-level localization even without GNSS/GPS signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If GPS technology is used for vehicle localization, then global position can be determined, but accuracy deteriorates in urban areas due to signal blocking
Solution Approach 1:
The patent introduces an intermediary system consisting of image sensors and dead reckoning algorithms that bridge the gap when GPS signals are blocked. The image sensors capture visual information of the environment, and dead reckoning processes this data to estimate vehicle position, serving as a mediator between the blocked GPS signal and the need for accurate localization.
Solution Approach 2:
The patent replaces the radio wave-based GPS system with an optical-based image sensing system for localization in urban areas. By substituting the electromagnetic wave propagation method (GPS) with optical image capture and processing, the system overcomes signal blocking issues caused by buildings and urban structures.
2Measurement precision
If high-definition maps are used to improve localization accuracy, then lane-level precision is achieved, but system cost increases
Solution Approach 1:
The patent employs standard-definition map data instead of expensive high-definition maps, and uses commercially available image sensors rather than specialized expensive equipment. By accepting that the localization estimate needs periodic updating through image processing, the system achieves cost-effective lane-level localization without requiring costly permanent infrastructure.
Solution Approach 2:
The patent changes the parameter of map resolution from high-definition to standard-definition, and adjusts the localization method from direct GPS to dead reckoning with image processing. This parameter change reduces system cost while maintaining acceptable localization accuracy through continuous image-based updates.
3Productivity
If dead reckoning is used to estimate vehicle location, then localization continues without GPS, but accuracy drifts over time
Solution Approach 1:
The patent implements feedback by continuously processing image data to update the vehicle's lateral position estimate. The image sensors provide ongoing visual feedback about the vehicle's environment and position relative to lane markings, allowing the dead reckoning system to correct drift and maintain accuracy over time rather than simply accumulating errors.
Data Source
AI summary
Embodiments relate to a method and apparatus for determining lane level localization of a vehicle with global coordinates. An automotive vehicle includes a vehicle navigation system, a database comprising standard-definition map data corresponding to road level latitude and longitude coordinates for a plurality of roadway features, at least one image sensor, and a processor in communication with the vehicle navigation system, the database, and the at least one image sensor. The processor is configured to determine an initial location of the vehicle, estimate a current location of the vehicle based on a dead reckoning process, detect, based on image data from the at least one image sensor, a lateral position of the vehicle within a lane of a road, and update the current location of the vehicle based on a difference between the estimated current location and the lateral position of the vehicle within the lane.