3D Point Cloud Shadow Analysis for Concealed Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Lidar systems face issues with crosstalk from highly reflective objects, causing concealed objects to be perceived larger than they actually are, leading to adjacent regions being overlit and objects behind a certain depth region being undetectable.
Innovation Solution
A method to determine concealed objects in a 3D point cloud by identifying shadows within a background region, using existing sensors to detect and evaluate shadows caused by crosstalk, without modifying hardware, and adding points to the 3D point cloud to represent concealed objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Length of stationary object
If Lidar systems use high sensitivity detection to measure distances up to 300m, then the detection range is improved, but crosstalk from highly reflective objects causes concealed objects to be perceived as larger and adjacent regions to be overlit
Solution Approach 1:
The system performs preliminary identification of shadow regions in the 3D point cloud before final object detection. By pre-identifying areas with insufficient point density that correspond to shadow zones caused by crosstalk, the system can compensate for the overestimation of object sizes and correct the measurement errors in advance.
Solution Approach 2:
The patent introduces shadow region identification as an intermediary step between raw Lidar data acquisition and final object detection. This intermediary process analyzes point cloud density and identifies shadow zones caused by crosstalk, allowing the system to correct measurement errors without modifying the fundamental Lidar hardware or detection range capabilities.
2Length of stationary object
If highly reflective objects are present in the environment, then the Lidar system can detect objects at longer distances, but crosstalk causes objects behind a certain depth region to be undetectable
Solution Approach 1:
The system preliminarily identifies shadow regions caused by crosstalk before attempting object detection. By pre-mapping the shadow zones where points are insufficient due to crosstalk interference, the system can reliably detect objects even behind highly reflective objects by compensating for the shadowed regions in advance.
Solution Approach 2:
The system uses feedback from shadow region identification to adjust object detection. When shadow zones are detected in the point cloud, the system modifies its detection algorithm to account for these regions, ensuring that objects behind highly reflective objects are not missed due to crosstalk interference.
3Productivity
If the Lidar system detects objects using standard processing methods, then the processing speed is maintained, but concealed objects are not revealed and measurement accuracy is reduced
Solution Approach 1:
The patent segments the 3D point cloud into distinct regions: shadow regions characterized by insufficient point density and non-shadow regions with adequate point density. This segmentation allows the system to process different regions with different algorithms, maintaining overall processing speed while improving measurement precision for concealed objects in shadow zones.
Solution Approach 2:
The system applies different processing qualities to different regions of the point cloud. Shadow regions receive specialized processing to identify and compensate for crosstalk effects, while non-shadow regions use standard processing methods. This local differentiation maintains processing efficiency while improving accuracy where needed.
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 the reliability and safety of environment sensing by revealing concealed objects, enhancing the accuracy of semi-autonomous or autonomous vehicles without requiring hardware modifications.
Implementation Method 1
An active sensor means a sensor that, in order to sense an environment, first emits a scanning signal into the environment (for example, sound waves or electromagnetic waves) and senses portions of the scanning signal backscattered by the environment to the active sensor
Implementation Method 2
Lidar systems are highly sensitive distance measuring devices which, on the basis of travel time measurements of transmitted and received laser light, determine a distance between a remote object in the environment of the lidar system and the lidar system
Implementation Method 3
In particular when highly reflective objects are present in the environment of the lidar system, it is possible for crosstalk resulting therefrom to cause such an object to be perceived as larger than it actually is by a detector of the lidar system
Data Source
AI summary
A method and a device for determining concealed objects in a 3D point cloud representing an environment. The method includes: producing a 3D point cloud using an active sensor, wherein each point of the 3D point cloud represents a distance measurement by the active sensor; determining a background region within the 3D point cloud; determining shadows within the background region of the 3D point cloud, shadows being the regions within the background region at which there are no points of the 3D point cloud; and determining, in the environment of the sensor, an object which is concealed by crosstalk in the sensor, by identifying, in the background region of the 3D point cloud, at least one shadow which cannot be attributed to an object in the foreground of the 3D point cloud.