Attention-Driven Landmark Detection for Autonomous Vehicle Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current GPS systems used in autonomous vehicles are not precise enough for accurate route planning, and direct correction methods require significant computing power, which can divert resources from autonomous driving functions.
Innovation Solution
An attention-driven landmark detection method using Convolutional Neural Networks (CNNs) to detect location information by integrating live feature maps with reference feature maps, applying tangent hyperbolic operations to refine sub-region coordinates, and utilizing High-Density Maps for accurate vehicle localization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS is used to localize the autonomous vehicle, then the vehicle can obtain location information, but the location accuracy is insufficient for route planning
Solution Approach 1:
The patent introduces landmark objects as intermediary elements between the GPS system and the route planning system. By detecting and matching landmarks in the environment with pre-stored landmark information, the system achieves precise location correction without directly modifying the GPS hardware or algorithms. The landmarks serve as mediators that bridge the gap between coarse GPS location data and the precise location information needed for route planning.
Solution Approach 2:
The patent replaces the purely signal-based GPS localization mechanism with a vision-based landmark detection and recognition system. Instead of relying solely on satellite signal processing, the system uses image processing and pattern recognition to detect landmarks, extract their features, and match them with stored landmark databases. This substitution of mechanical/signal-based system with vision-based system enables higher precision location determination.
2Measurement precision
If direct correction of GPS location information is applied to improve accuracy, then location precision increases, but computing power consumption increases significantly
Solution Approach 1:
The patent segments the image processing task into distinct stages: landmark detection, feature extraction, and matching with stored landmark information. By dividing the correction process into these modular segments, the system can efficiently allocate computing resources to each stage. The segmentation allows the system to process only relevant portions of the image and data, rather than performing exhaustive calculations on all image data, thus reducing overall computing power consumption while maintaining location accuracy.
Solution Approach 2:
The patent performs preliminary actions by pre-storing landmark information and characteristics in a database before the vehicle operates. During operation, the system only needs to detect and match landmarks against this pre-prepared database, rather than performing complete location calculations from scratch. This preliminary preparation of landmark data significantly reduces the real-time computing burden during vehicle operation, enabling accurate location correction with lower computing power consumption.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for detecting a location of a subject vehicle capable of an autonomous driving by using a landmark detection. And the method includes steps of: (a) a computing device, if a live feature map is acquired, detecting each of feature map coordinates on the live feature map per each of reference objects included in a subject data region corresponding to a location and a posture of the subject vehicle, by referring to (i) reference feature maps corresponding to the reference objects, and (ii) the live feature map; (b) the computing device detecting image coordinates of the reference objects on a live image by referring to the feature map coordinates; and (c) the computing device detecting an optimized subject coordinate of the subject vehicle by referring to 3-dimensional coordinates of the reference objects in a real world.