Lidar Reflection Clustering for False-Positive Detection Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Lidar sensors in vehicles often produce false-positive detections due to reflective materials like window panes, leading to potential dangerous interventions in automated driving systems, as they incorrectly measure distances and create phantom objects.
Innovation Solution
A method and device that check for laser pulses reflected multiple times at different distances, cluster initial reflections, and mark further reflections as false-positive if they occur on approximately homogeneous surfaces, particularly at larger distances, to differentiate between real and phantom targets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If lidar sensor scans all reflected laser pulses to detect objects, then detection coverage is improved, but false-positive detections increase due to reflective materials
Solution Approach 1:
The patent segments the detection process into distinct phases: identifying first reflections, clustering them to define regions, and then evaluating subsequent reflections against these regions. This segmentation allows the system to differentiate between valid objects and phantom targets caused by reflective materials like window panes.
Solution Approach 2:
The patent performs preliminary clustering of first reflections before evaluating further reflections. By pre-defining regions based on initial reflections and their spatial relationships, the system establishes a reference framework that enables rapid identification and filtering of false positives from reflective surfaces.
2Measurement precision
If lidar sensor marks all further reflections as potential targets, then detection sensitivity is improved, but reliability decreases due to phantom targets
Solution Approach 1:
The patent implements a feedback mechanism where the characteristics of first reflections (clustering, spatial distribution, distance relationships) are used to evaluate and validate subsequent reflections. This feedback loop enables the system to confirm or reject potential targets based on consistency with previously established reflection patterns, thereby improving both precision and reliability.
Solution Approach 2:
The patent changes evaluation parameters based on reflection order and spatial relationships. First reflections are clustered to establish baseline parameters (region boundaries, distance ranges), while further reflections are evaluated against these parameters with additional criteria (whether they fall within clustered regions, their distance relative to first reflections). This dynamic parameter adjustment improves measurement precision while filtering phantom targets.
3Measurement precision
If lidar sensor differentiates between various cluster types, then classification accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent applies local quality analysis by examining specific geometric properties of reflection clusters (planar surfaces, curved surfaces, extension in one or three spatial directions) rather than treating all clusters uniformly. This localized approach enables accurate classification of different target types (edges, dust/fog, volumetric objects) while keeping processing complexity manageable through focused feature extraction.
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 significantly increases the reliability of lidar sensor operations in real traffic scenarios, preventing false interventions and enhancing the safety of automated and autonomous driving systems by accurately distinguishing between real and phantom detections.
Implementation Method 1
for every laser pulse that is reflected back to the lidar sensor in a scan range of this, it is checked whether this is reflected back several times in different distances
Implementation Method 2
laser pulse that is reflected back several times
Data Source
AI summary
A method for determining false-positive detections of a lidar sensor in a scanning process of the surroundings of a vehicle involves, for every laser pulse that is reflected back to the lidar sensor in a scan range of this, checking whether this is reflected back several times in different distances. First reflections of a laser pulse that is reflected back several times are clustered and then, if a distance evaluation of reflections from the cluster produced reveals that the laser pulses are being reflected at a reflecting surface that is at least approximately homogeneous, further reflections following the respective first reflection are marked as false-positive detections.