Lidar Imaging Apparatus Using Polarization Camera for Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current lidar imaging systems for autonomous vehicles face challenges in accurately and efficiently extracting features in diverse environments, particularly due to the slow training process of deep learning algorithms and the need for restarting training with different sensor configurations and environmental conditions.
Innovation Solution
Incorporating a polarization camera and an electronic processing unit that utilizes polarization properties of light to enhance feature extraction, allowing for faster and more accurate data processing by combining lidar and polarization imaging, which includes using linear and circular polarizers and beam splitters to enable parallel measurements and data fusion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep learning algorithms are used for feature extraction, then object classification accuracy is improved, but training time and computational complexity increase significantly
Solution Approach 1:
The patent segments the imaging system into multiple independent photo detectors (first photo detector for intensity images, second photo detector for polarization images). This allows parallel acquisition of different types of image data simultaneously, reducing the time needed to collect training data and enabling more efficient feature extraction without requiring sequential processing of all data types.
Solution Approach 2:
The patent introduces a new dimension of measurement by adding polarization information to the traditional intensity-based imaging. The polarization camera captures additional data about the orientation and polarization state of light, providing extra features for machine learning algorithms to analyze. This dimensional expansion enables more accurate object classification without requiring a complete restart of training processes.
2Adaptability or versatility
If sensor configuration changes to adapt to different environmental conditions, then adaptability is improved, but training process must be restarted causing delays
Solution Approach 1:
The patent creates a universal imaging system where the same polarization camera and processing unit can handle multiple environmental conditions (intense sunlight, fog, rain) and various object types. The system uses beam splitters and polarizers to capture polarization information that remains valid across different lighting conditions, eliminating the need to restart training when environmental parameters change. The multi-functional sensor can adapt to different scenarios using existing training data.
3Device complexity
If conventional cameras are used for imaging, then device complexity is low, but feature extraction in diverse environments is insufficient
Solution Approach 1:
The patent merges conventional intensity-based imaging with polarization imaging in a single system. The beam splitter divides incoming light into separate paths that feed different photo detectors, combining multiple types of optical information acquisition. This integration allows the system to leverage both simple intensity data and sophisticated polarization data, improving feature extraction capabilities while maintaining manageable system complexity through modular design.
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 enables faster and more accurate feature extraction, improving data processing speed and algorithm learning by providing additional information on material types, which enhances object classification and detection in various environmental conditions, including intense sunlight, fog, and rain.
Implementation Method 1
said electronic processing unit is further adapted to perform feature extraction algorithms in image processing using polarization properties of light gained by said polarization camera
Implementation Method 2
using linear and circular polarizers and beam splitters to enable parallel measurements
Implementation Method 3
using linear and circular polarizers and beam splitters to enable parallel measurements
Implementation Method 4
said electronic processing unit is adapted to determine the time-of-flight of a laser pulse from said emitting section to said lidar detector
Implementation Method 5
each mirror of the DMD is controllable to change at least between a first position where light entering the imaging apparatus through said lens objective is directed on said lidar detector, and a second position where light entering the imaging apparatus through said lens objective is directed on said further photo detector
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
A lidar imaging apparatus (30) for a motor vehicle, comprising an emitting section (23) adapted to emit a pulsed light beam (3) and to change the direction of the light beam (3) in two dimensions perpendicular to the light beam (3), a receiving section (24) having a lens objective (11), a digital micromirror device (6) comprising a matrix of mirrors (7), a lidar detector (8) and a further photo detector (9), and an electronic processing unit (19), wherein each mirror (7) of the digital micromirror device (6) is controllable by said electronic processing unit (19) to change at least between a first position where light entering the lidar imaging apparatus (30) through said lens objective (11) is directed on said lidar detector (8), and a second position where light entering the lidar imaging apparatus (30) through said lens objective (11) is directed on said further photo detector (9), wherein said electronic processing unit (19) is adapted to determine the time-of-flight of a laser pulse from said emitting section (23) to said lidar detector (8), wherein the further photo detector (9) comprises a polarization camera (9) and said electronic processing unit (19) is further adapted to perform feature extraction algorithms in image processing using polarization properties of light gained by said polarization camera (9) .