LIDAR Vehicle Classification Using Rear Reflector Geometry
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous emergency braking systems face challenges in reliably classifying preceding vehicles due to reflection noise from color, geometric shape, and environmental factors, especially in bad weather, using LIDAR sensors.
Innovation Solution
A system and method that utilize the number and interval of rear reflectors to classify vehicles, employing a LIDAR sensor to generate image information on light reflection intensity, detect pixel areas corresponding to reflectors, and determine the operation point of the autonomous emergency braking system based on the classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If LIDAR sensor is used to detect preceding vehicle by amount of reflected light, then detection capability is improved, but classification reliability deteriorates due to reflection noise from color, geometric shape, and surrounding environment
Solution Approach 1:
The patent segments the reflected light detection into multiple wavelength bands (e.g., first wavelength band and second wavelength band). By dividing the detection into different spectral segments, the system can analyze reflection characteristics at different wavelengths, enabling more reliable vehicle classification while maintaining detection capability despite reflection noise from color, geometry, and environment.
2Length of stationary object
If RADAR sensor in milliliter wave band is used, then detection range is improved, but horizontal resolution deteriorates
Solution Approach 1:
The patent transitions from single-dimensional detection to multi-dimensional detection by incorporating wavelength band dimension. Instead of relying solely on spatial resolution in the horizontal dimension, the system adds spectral dimension through multi-band LIDAR detection, enabling accurate vehicle classification and type identification while maintaining adequate detection range.
3Illumination intensity
If stereo camera in visible ray area is used, then image quality is improved, but night environment detection capability deteriorates
Solution Approach 1:
The patent changes the detection parameter from visible light intensity to laser wavelength reflection characteristics. By using LIDAR sensors that emit laser beams at specific wavelengths and analyzing the reflected light properties rather than relying on ambient visible light, the system maintains detection and classification capability in low-light and night environments where stereo cameras fail.
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
Improves vehicle classification accuracy and minimizes erroneous detection by leveraging geometric characteristics of rear reflectors, allowing for more precise adjustment of the autonomous emergency braking system operation points, thereby reducing vehicle accident damages.
Implementation Method 1
uses a 3D image camera sensor, such as a radio detection and ranging (RADAR) sensor, a light detection and ranging (LIDAR) sensor
Implementation Method 2
collects laser beam reflected and returned from a front preceding vehicle
Data Source
AI summary
The present invention provides a system and a method of detecting a preceding vehicle. The system for detecting a preceding vehicle includes: an image sensor configured to generate image information containing information on reception intensity of light reflected from a preceding vehicle; a pixel detection unit configured to detect pixel areas corresponding to light reflected from rear reflectors of the preceding vehicle from the generated image information; and an Autonomous Emergency Braking (AEB) operation point controller configured to group adjacent pixel areas among the detected pixel areas, and classify the kind of vehicle of the preceding vehicle by using at least one element of information between information on the number of grouped pixel areas and information on an interval.