Vehicle Self-Localization Quality Estimation Under Landmark Occlusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous driving systems face challenges in accurately estimating future localization quality due to the visibility of environmental features, which is crucial for ensuring safe operation and switching between automated and manual control.
Innovation Solution
A method that predicts future landmark visibility by detecting static and dynamic obstacles, estimating their movement trajectories, and calculating the sensor's field of view, allowing for the determination of localization quality based on the number and distribution of observable landmarks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the vehicle uses environmental features (landmarks) for self-localization, then the accuracy of position determination is improved, but the reliability of localization deteriorates when landmarks are occluded by obstacles
Solution Approach 1:
The system performs preliminary actions by predicting future occlusions of landmarks before they occur. It detects static and dynamic obstacles, estimates their movement trajectories, and calculates when they will occlude landmarks, allowing the system to proactively assess localization quality and prepare for potential degradation before it happens.
Solution Approach 2:
The system applies beforehand cushioning by creating a buffer against potential localization failures. It estimates future localization quality by considering occlusions that will occur, and when the estimated quality falls below a threshold, it triggers a transition to manual control or activates backup localization methods, cushioning against the reliability deterioration that would occur if occlusions were not anticipated.
2Reliability
If the system transitions to manual control when localization quality is insufficient, then the safety is improved, but the productivity deteriorates due to loss of automated driving functionality
Solution Approach 1:
The system implements feedback by continuously monitoring the actual localization quality and comparing it with the estimated future localization quality. This feedback loop allows the system to make informed decisions about when to transition between automated and manual control, balancing safety requirements with maintaining automated driving functionality whenever possible.
Solution Approach 2:
The system applies dynamics by enabling flexible transition between automated and manual control modes based on real-time localization quality assessment. The control mode is not fixed but dynamically adjusts according to the estimated future localization quality, allowing the system to maintain automated driving when conditions permit and switch to manual control only when necessary for safety.
3Reliability
If the vehicle accounts for dynamic obstacles and their movement trajectories, then the reliability of localization quality estimation is improved, but the device complexity increases
Solution Approach 1:
The system applies universality by using a multi-functional sensor device that performs both obstacle detection and localization quality estimation. The same sensor system that detects obstacles for collision avoidance is also used to predict future occlusions of landmarks, eliminating the need for separate dedicated hardware and reducing overall system complexity despite the enhanced functionality.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method for estimating the quality of localisation, which can include the following steps: Using sensor detection, the vehicle (10) detects dynamic objects on the road and in the direct surroundings of the road, for example other vehicles, and estimates their dimensions. The movement of these dynamic objects (12) in the near future is estimated. The outer casings of these objects (12) are entered into a map of the surroundings (UK). From the perspective of the sensors (150, 186) used to detect the features in the surroundings, the limitations of the fields of view and the predicted temporal development thereof as the result of the movement of the vehicle (10) itself and the predicted movements of the dynamic objects (12) are entered into the map of the surroundings (UK). The surrounding features that have been entered into the map of the surroundings (UK) and which may at best be visible in the near future are determined. An upper limit for a measure of the quality of localisation is estimated.