LiDAR Computer Unit Halation Filtering for Autonomous Vehicles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
LiDAR devices face issues with halation, which distorts measurements due to crosstalk from highly reflective objects, potentially leading to overestimation of object sizes and triggering false emergency braking in autonomous vehicles, compromising road safety.
Innovation Solution
A computer unit filters out halation from laser signal data points by identifying and separating reflective and real objects using algorithms that analyze intensity levels, echo durations, and statistical noise to accurately distinguish between retroreflectors and real objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If LiDAR devices use parallelized measurement (flash LiDAR, vertical flash LiDAR, or horizontal flash LiDAR), then measurement efficiency and speed are improved, but halation distortion becomes more pronounced
Solution Approach 1:
The patent applies preliminary action by performing halation filtering before object identification and classification. The computer unit first identifies and removes halation-affected data points from the point cloud, then proceeds with object detection. This preliminary removal of distorted data prevents halation from interfering with subsequent measurement and object recognition processes, thereby maintaining measurement precision while preserving the high productivity of parallelized LiDAR measurement.
2Measurement precision
If halation filtering is applied to remove distorted data points, then measurement precision is improved, but risk of removing valid object data points increases
Solution Approach 1:
The patent applies local quality by implementing spatially varying filtering thresholds and intensity analysis. Instead of uniform filtering across the entire field of view, the system analyzes local intensity distributions and applies adaptive filtering only where halation is detected. The computer unit examines intensity levels of individual data points and their neighbors, applying filtering selectively to regions with halation characteristics while preserving data points in regions with valid object reflections. This localized approach maintains measurement precision by removing halation while preserving object detection reliability by retaining valid data points.
Solution Approach 2:
The patent implements feedback through iterative validation and cross-checking of filtered data points. The computer unit uses multiple criteria (intensity thresholds, spatial distribution patterns, temporal consistency) to validate whether data points should be retained or removed. The system continuously adjusts filtering parameters based on the detected patterns in the point cloud, providing feedback loops that prevent over-filtering. This feedback mechanism ensures that halation is effectively removed while maintaining high reliability in object detection by preserving genuine object data points.
3Measurement precision
If halation is filtered out using intensity level thresholds, then halation removal effectiveness is improved, but false identification of real objects increases
Solution Approach 1:
The patent applies another dimension by extending the filtering criteria from single-dimensional intensity thresholding to multi-dimensional analysis. The computer unit evaluates data points based on multiple parameters simultaneously: intensity level, spatial position, temporal characteristics, and local neighborhood patterns. By adding these additional dimensions to the filtering process, the system achieves more effective halation removal while reducing false identification of real objects. The multi-criteria approach allows differentiation between halation-affected points and valid object points even when their intensity levels overlap, thereby improving both filtering effectiveness and object identification accuracy.
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
The solution effectively filters out halation, allowing for reliable identification of real objects near retroreflectors, preventing false safety triggers and ensuring accurate detection of gantries and objects below them, thereby enhancing road safety for autonomous vehicles.
Implementation Method 1
a laser source configured to emit a laser signal into a transmit path
Implementation Method 2
Said laser signal is reflected on objects in the surroundings of the LiDAR device. The reflected laser signal impinges on the LiDAR device again
Implementation Method 3
The LiDAR sensor then calculates how far away the object is from the LiDAR device using the measured time-of-flight difference
Implementation Method 4
If the object is highly reflective, and in particular retroreflective, crosstalk with adjacent laser signal data points may occur. This is referred to as halation
Data Source
AI summary
A computer unit for a LiDAR device, which has a laser source configured to emit a laser signal into a transmit path, and a LiDAR sensor arranged in a receive path and configured to detect a laser signal reflected into the receive path. The computer unit is configured to process a multiplicity of laser signal data points of the reflected laser signal. The computer unit is configured to filter halation out of the laser signal data points of the reflected laser signal.