Dual LiDAR Obstacle Detection Ground Estimation Blind Zone
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In autonomous driving, laser radars with a high number of wiring harnesses create a detection blind zone near the ground due to their mounting height and angle, leading to incomplete obstacle detection and potential false detections caused by vehicle vibration or inaccurate mounting.
Innovation Solution
The method involves acquiring point cloud data from both a high-mounted laser radar with many wiring harnesses and a low-mounted laser radar, performing ground estimation on the first data set, filtering out ground points from the second data set, and then fusing both data sets for comprehensive obstacle detection, thereby reducing false positives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a laser radar with a high number of wiring harnesses is mounted on the vehicle body at a certain height, then the laser radar can perform obstacle detection, but a detection blind zone is formed in the area close to the ground
Solution Approach 1:
The system divides the obstacle detection task into two segments: the first laser radar (high-mounted) detects obstacles in the upper and middle regions, while the second laser radar (low-mounted) detects obstacles in the lower region near the ground. This segmentation allows each radar to operate in its optimal detection zone without interference from wiring harnesses, eliminating the blind zone while maintaining overall detection precision.
Solution Approach 2:
The system introduces a vertical dimension differentiation by mounting the second laser radar at a lower height than the first laser radar. This dimensional change allows the second radar to detect obstacles in the previously blind lower region near the ground, complementing the first radar's detection coverage and eliminating the detection blind zone.
2Stability of the object's composition
If the laser radar is mounted at a certain height with wiring harnesses, then the structure is stable, but vehicle vibration causes false detections
Solution Approach 1:
The system introduces ground estimation as an intermediary process that mediates between the raw point cloud data and obstacle detection results. By estimating the ground surface and filtering points based on their distance from the estimated ground, the system eliminates false detections caused by vehicle vibration while maintaining stable mounting structure.
Solution Approach 2:
The system implements a feedback mechanism where the ground estimation result is used to filter points in the second point cloud data. This feedback loop continuously refines the detection by comparing actual points against the estimated ground model, removing false detections caused by vibration while preserving true obstacles.
3Productivity
If ground points are not filtered from the second point cloud data, then all detected points are processed, but false detections increase due to vehicle vibration
Solution Approach 1:
The system extracts and removes ground points from the second point cloud data by calculating the distance between each point and the estimated ground surface. Points within a threshold distance are identified as ground points and filtered out, eliminating the source of false detections while preserving all other points for obstacle detection.
Solution Approach 2:
The system performs ground estimation and ground point filtering as preliminary actions before final obstacle detection. By pre-processing the second point cloud data to remove ground points, the system reduces the computational load during obstacle detection and eliminates false detections caused by vehicle vibration in advance.
Data Source
AI summary
Embodiments of the present disclosure provide a method and apparatus for detecting an obstacle. The method may include: acquiring first point cloud data collected by a first vehicle-mounted laser radar and second point cloud data collected by a second vehicle-mounted laser radar, where a height of the first vehicle-mounted laser radar from a ground is greater than a height of the second vehicle-mounted laser radar from the ground, and a number of wiring harnesses of the first vehicle-mounted laser radar is greater than a number of wiring harnesses of the second vehicle-mounted laser radar; performing ground estimation based on the first point cloud data; filtering out a ground point in the second point cloud data according to the ground estimation result of the first point cloud data; and performing obstacle detection based on the second point cloud data after the ground point is filtered out.