Vehicle Self-Position Estimation Using Adaptive Point Group Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing self-position estimation systems for autonomous driving and driving support systems face inaccuracies due to the lack of consideration for camera point group observation accuracy and local adaptation, leading to incorrect or low-accuracy self-position estimates, especially in residential areas with limited detailed maps.
Innovation Solution
An on-board processing device that includes a past map storage unit, a self-map generation unit, and a data selection unit, which generates a traveling map from sensor data and selects optimal data for self-position estimation based on the characteristics of the traveling map, considering the accuracy of the point group and local adaptation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If self-position estimation is performed using camera point groups without considering observation accuracy, then the system can operate in areas without detailed maps, but the self-position estimation accuracy becomes low
Solution Approach 1:
The system dynamically changes parameters by selecting different point groups based on observation accuracy conditions. When observation accuracy is high, it uses camera point groups for high adaptability; when accuracy is low, it switches to alternative point groups or methods to maintain estimation accuracy
Solution Approach 2:
The system implements dynamic selection of point groups based on real-time observation accuracy assessment. The self-position estimation unit adaptively switches between different point group sources (camera-based or alternative) depending on current conditions, making the system both accurate and versatile
2Reliability
If weather information is considered for self-position estimation, then reliability improves, but camera point group observation accuracy and local adaptation are not considered leading to estimation errors
Solution Approach 1:
The system segments the point group selection process into multiple independent sources (camera point groups, alternative point groups, weather-based corrections). By dividing the estimation task into separate evaluable components, it can assess observation accuracy for each segment and combine them appropriately to achieve both reliability and precision
3Ease of manufacture
If detailed maps are prepared only for expressways, then map preparation cost is reduced, but self-position estimation accuracy deteriorates in residential areas
Solution Approach 1:
The system performs self-service by generating its own point groups from camera observations when pre-prepared detailed maps are unavailable. This allows the vehicle to create and use its own reference data in residential areas, maintaining estimation accuracy without requiring external map preparation resources
Data Source
AI summary
An on-board processing device includes: a past map storage unit that stores a past generated map generated in the past; a self-map generation unit that generates a traveling map on the basis of sensor data acquired while traveling; a self-position estimation unit that estimates a self-position by collating the past generated map and the traveling map; and a data selection unit that selects data, to be used by the self-position estimation unit, from the traveling map on the basis of characteristics of data included in the traveling map.