LiDAR Object Detection via Point Cloud Absence Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection systems based on three-dimensional point clouds fail to detect objects when the point cloud associated with the object is not generated, such as with metal objects or black bodies that specularly or poorly reflect laser light.
Innovation Solution
The system includes a point cloud missing region extraction unit that identifies regions where a point cloud is missing and an object estimation unit that estimates the presence of an object based on these missing regions, allowing detection even without a complete point cloud.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LiDAR apparatus is used to generate three-dimensional point cloud, then ranging capability is improved, but detection capability for metal objects and black bodies deteriorates
Solution Approach 1:
The patent introduces an intermediary inference mechanism that uses surrounding point cloud data to detect objects that do not reflect laser light. Instead of directly detecting the metal object or black body through laser reflection, the system infers their presence by analyzing the absence of point clouds in regions where objects should exist based on environmental context and physical laws, thus resolving the contradiction between maintaining ranging precision and improving detection reliability for non-reflective objects
Solution Approach 2:
The patent converts the harmful effect of laser light being specularly reflected or absorbed (which causes detection failure) into a beneficial inference opportunity. By recognizing that the absence of point clouds in certain regions indicates the presence of non-reflective objects, the system transforms the detection problem into an inference problem, improving reliability without sacrificing the original ranging capability
2Measurement precision
If object detection is performed based on three-dimensional point cloud, then detection accuracy is improved, but detection capability for objects without point cloud deteriorates
Solution Approach 1:
The patent makes the object detection system universal by enabling it to handle both traditional point cloud-based detection and inference-based detection for objects without point clouds. The system performs multiple functions: direct object detection from point clouds, inference of objects from missing point clouds, and integration of both approaches, thereby improving detection versatility while maintaining accuracy for detectable objects
3Quantity of substance
If laser light is emitted for ranging, then three-dimensional point cloud generation is improved, but point cloud acquisition for specularly reflecting objects deteriorates
Solution Approach 1:
The patent applies preliminary action by inferring the presence of objects before direct detection fails. By analyzing the environment and predicting where objects should exist based on physical context, the system prepares inference results in advance that can compensate for the loss of point cloud information from specularly reflecting objects, thus reducing information loss while maintaining adequate point cloud data for other purposes
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 the detection of objects in scenarios where a complete three-dimensional point cloud is not available, improving the system's ability to identify objects that would otherwise be undetectable.
Implementation Method 1
the LiDAR apparatus is a ranging apparatus on an assumption that the laser light emitted toward the ranging target is reflected by a surface of the ranging target and is input to the LiDAR apparatus again
Data Source
AI summary
An object detection system includes: a point cloud missing region extraction means for extracting a point cloud missing region being a region where a point cloud is missing in a three-dimensional point cloud; and an object estimation means for estimating presence of an object, based on the point cloud missing region.