Vehicle Self-Position Estimation Using Reliable Road Targets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing self-position estimation methods for autonomous vehicles are prone to errors due to calibration issues and unstable estimation results when detecting white lines, leading to reduced accuracy.
Innovation Solution
A self-position estimation method and device that utilizes a combination of Laser Range Finders, cameras, and vehicle sensors to detect and process surrounding environment data, including white lines and curbs, while eliminating target position data with high errors by assessing reliability based on distance, attribute, continuous detection, and error distribution, thereby improving estimation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If white lines are used for self-position estimation, then the estimation process can be simplified, but the estimation accuracy is reduced due to calibration errors and position offsets
Solution Approach 1:
The patent segments the target objects into multiple categories (white lines, curbs, road signs, buildings) and processes their position data separately. By dividing the estimation process into multiple independent detection channels, the system can selectively weight or exclude data from specific sources (like white lines with calibration errors) while maintaining simplicity in the overall estimation framework.
Solution Approach 2:
The patent applies local quality by assigning different reliability weights to different target objects based on their individual characteristics and detection quality. Instead of treating all targets uniformly, the system evaluates each target's position data quality locally and adjusts its contribution to the final self-position estimation, thereby improving accuracy without significantly increasing process complexity.
2Reliability
If multiple target objects are used for estimation, then the reliability of position data can be improved, but the complexity of processing increases
Solution Approach 1:
The patent implements feedback mechanisms where the system continuously evaluates the quality and reliability of position data from multiple targets, and adjusts the weighting or selection of targets based on this feedback. This allows the system to dynamically optimize which targets to use for estimation, improving reliability while keeping processing complexity manageable through adaptive rather than exhaustive processing.
Solution Approach 2:
The patent changes parameters such as the reliability weights and selection criteria for different target objects based on their detection quality, environmental conditions, and calibration status. By dynamically adjusting these parameters, the system can incorporate multiple targets to improve reliability without requiring equally complex processing for all targets at all times.
3Quantity of substance
If all detected target position data is used for estimation, then more data is available for calculation, but errors from misaligned targets reduce accuracy
Solution Approach 1:
The patent extracts and removes target position data that is estimated to contain significant errors, such as white lines with calibration offsets or targets with poor detection quality. By selectively extracting only the reliable position data from the full set of detected targets, the system maintains sufficient data quantity for accurate estimation while eliminating the harmful effect of erroneous data points.
Solution Approach 2:
The patent converts the potential harm of having erroneous target data into a benefit by using the error detection and elimination process as a quality control mechanism. The system identifies targets with calibration errors or position offsets and either corrects them or excludes them, thereby transforming the challenge of data quantity into an opportunity for improved data quality through systematic error filtering.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The method enhances the accuracy of self-position estimation by selectively using reliable target position data, reducing errors and stabilizing the estimation process.
Implementation Method 1
a result (surrounding environment information) of having detected a movable region of a mobile robot by means of a sensor
Implementation Method 2
both white lines are simultaneously detected
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A self-position estimation method includes: detecting a relative position between a target present in surroundings of a moving object and the moving object; storing a position where the relative position is moved by the moved amount of the moving object, as target position data; selecting the target position data on the basis of reliability of the relative position of the target position data with respect to the moving object; and comparing the selected target position data with map information including the position information on the target present on a road or around the road, thereby estimating a self-position which is a current position of the moving object.