LiDAR Object Outline Segments With Confidence Scoring for Map Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional LiDAR-based object detection techniques lack reliability assessment for segments, leading to poor map-matching and vehicle localization results due to the use of low-reliability segments, particularly in cases where segments are long, have missing representative points, or are not vertically overlapped in the point cloud.
Innovation Solution
A processor determines confidence scores for segments in a LiDAR-based object detection system by evaluating conditions such as segment length, angle, and presence of representative points, assigning scores of 0 or 1 to segments based on predetermined criteria, and excluding low-confidence segments from the map-matching process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If segments are used for map-matching without reliability assessment, then the map-matching process can be performed, but the reliability of map-matching results and vehicle localization deteriorates
Solution Approach 1:
The patent applies preliminary action by calculating confidence scores for all segments before the map-matching process. The processor determines confidence scores based on geometric conditions (segment length, angle, presence of representative points) in advance, then uses these scores to filter segments. This ensures that only reliable segments are used in map-matching, preventing reliability deterioration while maintaining process efficiency.
Solution Approach 2:
The patent applies local quality by assigning different confidence scores to different segments based on their individual characteristics. Each segment is evaluated independently using conditions such as segment length, angle with adjacent segments, and presence of representative points. This allows the system to selectively use high-confidence segments while excluding low-confidence ones, improving overall map-matching reliability without compromising efficiency.
2Productivity
If all segments are included in object detection, then the detection process is complete, but the accuracy of object outline detection deteriorates due to low-reliability segments
Solution Approach 1:
The patent applies parameter changes by introducing confidence scores as a new parameter to evaluate segment reliability. The processor calculates confidence scores based on geometric parameters (segment length, angle, representative point distribution) and uses these scores to filter segments. This allows the system to maintain detection completeness while improving outline detection accuracy by excluding segments that fail to meet reliability thresholds.
3Reliability
If confidence score calculation with multiple conditions is implemented, then the reliability of detection results improves, but the device complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the confidence score calculation into distinct, independent conditions: segment length condition, angle condition, and representative point presence condition. Each condition is evaluated separately and contributes to the overall confidence score. This modular approach improves detection reliability through comprehensive evaluation while managing processor complexity by organizing the calculation into discrete, manageable segments.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances the reliability of object detection and autonomous driving by improving the accuracy of map-matching and vehicle localization by filtering out unreliable segments, thereby improving overall autonomous driving performance.
Implementation Method 1
a Light Detection And Ranging (LiDAR) sensor configured to obtain a point cloud
Data Source
AI summary
A Light Detection And Ranging (LiDAR)-based object detection apparatus comprising a LiDAR sensor configured to obtain a point cloud, and a processor configured to detect at least one object of interest from the point cloud, wherein the processor is configured to perform determining representative points from LiDAR points corresponding to the object among the point cloud, determining outer points among the representative points, the outer points defining an outline of the object, and determining a confidence score for each of segments connecting at least two of the outer points.