3D Point Cloud Ground Identification Without Pre-Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in accurately identifying the ground surface in three-dimensional point clouds without pre-mapping processes, leading to increased computational requirements and errors due to the need for extensive mapping and resource allocation.
Innovation Solution
A method involving generating a grid-based ground information system using a processor to identify and exclude the static object area and exclude the static object area from the calculation, thereby reducing errors and improving computational efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If pre-mapping processes are used to identify ground in point clouds, then ground identification accuracy is improved, but computational resources and time are increased
Solution Approach 1:
The patent divides the three-dimensional space into multiple two-dimensional planes at different height levels. Each plane is processed independently to identify ground points, allowing parallel computation and reducing overall processing time while maintaining identification accuracy.
Solution Approach 2:
The patent transforms the traditional three-dimensional point cloud processing into multiple two-dimensional plane processing tasks. By projecting points onto 2D planes at different heights and processing them separately, the computational complexity is reduced while preserving ground identification precision.
2Reliability
If extensive mapping is performed to identify ground area, then ground identification completeness is improved, but computational requirements are increased
Solution Approach 1:
The patent processes only the necessary portions of the point cloud by dividing space into multiple planes and processing each plane independently. This partial processing approach ensures complete ground identification without the need to process the entire 3D space as a single unit, reducing computational requirements.
Solution Approach 2:
By segmenting the 3D space into multiple 2D planes, the patent enables independent processing of each plane's ground identification task. This segmentation ensures that all ground areas are covered completely while distributing computational load across multiple smaller tasks.
3Area of stationary object
If multiple sensing devices are deployed to cover wide spaces, then sensing coverage is improved, but system complexity and resource allocation are increased
Solution Approach 1:
The patent merges data from multiple sensing devices by projecting their point clouds onto the same set of 2D planes. This unified processing approach maintains wide area coverage while simplifying system management by using a common processing framework for all devices.
Solution Approach 2:
The 2D plane processing framework serves as a universal method that can handle data from any number of sensing devices. Each device's data is processed through the same plane-based algorithm, making the system scalable and reducing complexity associated with device-specific processing.
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
Accurately identifies the ground surface in three-dimensional point clouds without pre-mapping, reducing computational resources and time, and enhancing computational efficiency by distinguishing between static and dynamic objects.
Implementation Method 1
LiDAR (Light Detection And Ranging), one of the 3D sensors, is a sensor that obtains information about 3D space by emitting light to an object in 3D space and then receiving the reflected light
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An electronic device includes a communication circuit, an input device, at least one processor, and at least one memory. The at least one processor stores instructions, and the instructions are configured to obtain a time-series point cloud for a three-dimensional space from the sensing device, to generate a grid covering the three-dimensional space and having a plurality of cells, to select at least on cell having a ground among the plurality of the cells through the input device, to determine a height value of each of the at least one cell based on the point cloud, to determine a height value of each of remaining cells excluding the at least one cell among the plurality of the cells based on the height value of each of the at least one cell, and to identify the ground based on the height value of each of the plurality of the cells.