Autonomous Vehicle Localization Error Correction via Real-Time Landmark Deviation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in high precision localization, especially on road segments with few, small, or unclear landmarks, leading to inaccuracies in localization based on outdated HD maps.
Innovation Solution
Implementing a real-time localization error correction system using autonomous vehicles to detect ground truth landmarks and calculate deviations from reference landmarks stored in a HD map experience management system, allowing for real-time corrections and updates to improve localization accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the AV uses landmarks from outdated HD maps for localization, then the system can operate with pre-stored map data, but localization accuracy deteriorates on road segments with few, small, or unclear landmarks
Solution Approach 1:
The system continuously compares detected landmarks with HD map landmarks and uses the deviation information to generate correction values. These correction values are fed back to update the localization results in real-time, creating a feedback loop that maintains high accuracy despite map outdatedness
Solution Approach 2:
The system pre-stores correction values for different road segments and applies them proactively when the AV traverses those segments, before the outdated map data would cause significant localization errors
2Measurement precision
If the AV relies on pre-stored HD map landmarks, then the localization system has lower computational requirements, but it cannot adapt to real-time changes or deviations in landmark positions
Solution Approach 1:
The AV system performs self-correction by automatically detecting its own localization deviations and generating correction values without external intervention, making the system self-sufficient and reducing dependency on complex external correction services
Solution Approach 2:
The system changes the parameter state by transforming static HD map landmark data into dynamic correction values that are updated in real-time based on actual detected deviations, allowing the system to adapt to changing conditions
3Reliability
If the system uses real-time detected landmarks to correct localization errors, then localization accuracy improves on road segments with limited landmarks, but computational processing requirements increase
Solution Approach 1:
The system applies partial correction by using real-time detection only when necessary (when landmarks are detectable) and partial correction values (deviation from reference), rather than continuously processing all possible landmarks, thus reducing computational burden while maintaining reliability where needed
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure relates to real-time localization error correction of an autonomous vehicle (AV). A processor for real-time localization error correction of the AV is provided. The processor is configured to retrieve a reference landmark around the AV from a map aggregating server (MAS), wherein the AV is configured to interact with the MAS for real-time localization; detect, in real time, a ground truth landmark corresponding to the reference landmark, according to image data captured by one or more image capture devices installed on the AV; and determine a deviation between the ground truth landmark and the reference landmark as a real-time correction value for the real-time localization of the AV.