2D Pixel Overlap Detection for Lane and Object Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current driver assistance systems face errors in identifying lanes and selecting target vehicles due to ambiguous and noisy transformations from two-dimensional to three-dimensional space, especially at greater distances, leading to incorrect object selection and potential failure in automated vehicle interventions.
Innovation Solution
A method that determines the degree of overlap between objects and lanes using a two-dimensional pixel data field, where lane and object pixel groups are compared directly without transformation, allowing for accurate allocation and identification of objects relative to lanes, thereby improving the accuracy of lane changes and occupancy recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If lane markings are transformed from two-dimensional image space into three-dimensional measurement space, then lane information can be integrated with radar data, but measurement precision deteriorates due to transformation ambiguity and noise
Solution Approach 1:
Instead of transforming lane markings from 2D image space into 3D measurement space (conventional approach), the patent inverts the approach by keeping lane markings in 2D pixel data field and transforming radar object data into the 2D pixel data field for comparison. This avoids the ambiguous and noisy transformation from 2D to 3D space while achieving the same integration goal.
2Device complexity
If video-based lane marking recognition is used in a one-sensor strategy, then device complexity is reduced, but measurement precision deteriorates due to exponential noise increase with distance
Solution Approach 1:
The patent creates a 2D representation (copy) of radar object data in the pixel data field that matches the format of lane markings. This copying approach allows direct comparison in 2D space without requiring complex 3D transformations, maintaining precision at distance while keeping the one-sensor strategy simple.
3Ease of operation
If transformation from two-dimensional to three-dimensional space is performed, then objects can be selected in three-dimensional space, but reliability deteriorates due to incorrectly identified lanes
Solution Approach 1:
The patent inverts the conventional transformation direction by keeping lane markings in 2D pixel space and projecting radar objects into the same 2D pixel space. This eliminates the source of errors (2D to 3D transformation) while maintaining the ability to select target vehicles through overlap determination in the 2D pixel data field.
Data Source
AI summary
A method for determining a degree of overlap of at least one object with at least one lane via a representation of a surrounding environment of a platform as a two-dimensional pixel data field. The method includes allocating at least one lane pixel group to pixels of the two-dimensional pixel data field, which correspondingly represent at least one lane; allocating at least one object pixel group to pixels of the two-dimensional pixel data field, which correspondingly represent at least one object; defining at least one object pixel pair in the two-dimensional pixel data field, which pair characterizes a width of the at least one object pixel group; and comparing the object pixel pair and the lane pixel group in the two-dimensional pixel data field.