Vehicle Self-Position Estimation Using Dynamic Reference Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The accuracy of matching between registration images and observation images is reduced due to significantly different environments such as weather, time of day, and presence/absence of movable bodies, which in turn affects the accuracy of estimating the self-position of a vehicle.
Innovation Solution
A computerized method and apparatus for estimating the position of a movable object, including a self-position estimation unit that processes data from various sensors to generate a key frame map, invalidate movable body areas, and adjust 3D models based on weather information, improving matching accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If matching is performed between registration images and observation images captured under significantly different environmental conditions, then the system can operate in diverse weather and time conditions, but the matching accuracy deteriorates
Solution Approach 1:
The patent segments the image matching process into multiple stages: initial matching under current environmental conditions, followed by iterative refinement using weather and time information. The matching process is divided into coarse alignment and fine adjustment phases, allowing the system to handle environmental variations systematically while maintaining accuracy.
Solution Approach 2:
The patent applies preliminary actions by pre-processing images with weather correction filters and time-based adjustments before matching. Environmental parameters such as weather conditions and time of day are incorporated into the matching algorithm in advance, allowing the system to compensate for environmental differences before they degrade matching accuracy.
2Measurement precision
If multiple environmental factors (weather, time, movable bodies) are considered in image matching, then the estimation accuracy of self-position improves, but the computational complexity increases
Solution Approach 1:
The patent applies local quality by focusing computational resources on the most influential environmental factors for each specific matching scenario. Rather than uniformly processing all environmental data, the system selectively applies weather corrections, time adjustments, and movable body compensations based on their relative importance in the current context, reducing overall complexity while maintaining accuracy.
Solution Approach 2:
The patent implements dynamics by making the matching process adaptive and iterative. The system dynamically adjusts the level of environmental correction applied based on the degree of environmental difference detected between registration and observation images. This allows the computational complexity to scale with the actual need, rather than operating at maximum complexity continuously.
3Measurement precision
If environmental corrections are applied to registration images, then the matching accuracy under different weather conditions improves, but the processing time increases
Solution Approach 1:
The patent applies periodic action by implementing environmental corrections at strategically selected intervals rather than continuously. The system performs weather and time-based adjustments periodically during the matching process, particularly at critical stages such as initial alignment and major refinement points, rather than applying corrections at every processing step. This reduces cumulative processing time while maintaining the accuracy benefits of environmental adaptation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
There is provided methods and apparatus for estimating a position of a movable object. A dynamic reference image is generated based on an environment and a reference image extracted from a map. A position of the movable object is estimated based on the dynamic reference image and an observation image of an area around the movable object.