Object-to-Lane Association via ML Coordinates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques struggle to associate objects adjacent to a road, such as signs, with the corresponding lanes, which is crucial for providing accurate navigational information and controlling autonomous vehicles.
Innovation Solution
A method and apparatus that use a trained machine-learning model to determine the coordinates of object-to-lane association points, allowing objects to be accurately associated with the lanes they pertain to, even if they are positioned beside or above the road.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If bounding boxes and lane boundaries are used to determine vehicle lanes, then vehicle lane identification is possible, but objects adjacent to the road (such as signs) cannot be accurately associated with lanes
Solution Approach 1:
The patent segments the association task into two parts: first identifying objects and lanes separately using detectors, then establishing connections between them through association points. This allows the system to handle both vehicles and road signs with lane association capability.
Solution Approach 2:
The patent introduces association points as intermediary elements that connect objects to lanes. These association points serve as mediators between object detection results and lane detection results, enabling accurate association without requiring direct spatial overlap between objects and lanes.
2Device complexity
If traditional object detection and lane detection are performed separately, then detection simplicity is maintained, but association accuracy between objects and lanes deteriorates
Solution Approach 1:
The patent merges object detection, lane detection, and association determination into a unified processing pipeline. By combining these functions and using association points to link objects to lanes, the system achieves both simplicity in detection process and accuracy in association without requiring separate complex modules.
Data Source
AI summary
Systems and techniques are described herein for determining lane information. For instance, a method for determining lane information is provided. The method may include obtaining an image representative of one or more lanes of a road and an object, wherein the object is adjacent to the road; and determining coordinates of object-to-lane association points of at least one lane of the one or more lanes of the road, wherein the coordinates are associated with the object