Coaxial Scanning Lidar Multipath Interference Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing LIDAR systems face challenges in accurately measuring distances due to multipath interference, where multiple reflections of laser light from surrounding surfaces interfere with the direct reflection from the object, leading to errors in depth mapping.
Innovation Solution
A simulation method and system for coaxial scanning LIDAR that models and combines reflection and multiple interference light paths to simulate mixed light, allowing for the removal of multipath interference through trained extreme gradient boosting (XG Boost) algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LIDAR measures distance using reflected light from objects, then distance measurement is achieved, but multipath interference from multiple reflections causes measurement errors
Solution Approach 1:
The patent converts the harmful multipath interference into a beneficial signal by using the same multiple reflections that cause interference to provide additional information about the environment. The system models and simulates these multiple reflection paths to create training data that helps the machine learning algorithm distinguish between direct and reflected light paths, ultimately using the interference patterns themselves to improve measurement accuracy.
Solution Approach 2:
The patent introduces a machine learning algorithm as an intermediary between the raw LIDAR signals and the distance measurement process. This algorithm acts as a mediator that processes the mixed signals containing both direct reflection and multipath interference, separating and identifying the direct reflection component to enable accurate distance measurement while filtering out interference.
2Measurement precision
If existing methods correct multipath interference, then measurement accuracy improves, but correction takes too much time for real-time application
Solution Approach 1:
The patent performs preliminary actions by pre-simulating multiple reflection paths and pre-training the machine learning algorithm with this simulated data before actual real-time measurement. This offline preparation creates a ready-to-use model that can quickly process real LIDAR signals without requiring complex real-time calculations, thus achieving both accuracy and speed.
Solution Approach 2:
The patent creates copies of the complex simulation environment to generate training data. By simulating various multipath scenarios and creating corresponding labeled training datasets, the system prepares the machine learning model in advance, allowing rapid real-time inference without repeating the complex simulations during actual measurement.
3Loss of information
If LIDAR receives mixed light including direct reflection and multiple interference light, then comprehensive environmental information is captured, but distance measurement errors occur
Solution Approach 1:
The patent segments the mixed light signal into distinct components by using a machine learning algorithm to identify and separate direct reflection light from multipath interference. The system analyzes signal characteristics such as intensity, timing, and pattern recognition to divide the composite signal into separable elements, allowing accurate identification of the direct reflection component for distance measurement while preserving awareness of environmental features from reflected paths.
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 proposed solution enables accurate estimation and removal of multipath interference, resulting in improved depth map accuracy and real-time correction of distortion caused by multiple interference lights.
Implementation Method 1
an avalanche photodiode that detects a mixed light in which reflection light directly reflected from the object and multiple interference light are mixed
Implementation Method 2
a laser that transmits measurement light
Implementation Method 3
a two-axis scanner that scans an object by rotating the measurement light about a coaxial axis
Data Source
AI summary
According to one embodiment of the present disclosure, a multipath interference correction system for a coaxial scanning LIDAR, includes a scanner driving unit configured to drive a scanner included in a coaxial scanning LIDAR and obtain a position signal of the scanner, a measurement light modulation signal input unit configured to obtain a demodulation signal of measurement light emitted from laser of the coaxial scanning LIDAR, a detection light input unit configured to receive detection light mixed with reflection light and multiple interference light from an avalanche photodiode of the coaxial scanning LIDAR and output a detection light signal, and a real-time processor for outputting a depth map based on the detection light signal, the demodulation signal of the measurement light, and the position signal of the scanner.