3D Sensor Environment Mapping by Farthest-Point Landmark Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional SLAM techniques face challenges in generating accurate environment maps in environments with moving objects due to insufficient point-cloud matching, high processing loads, and incorrect landmark recognition of moving objects as static features.
Innovation Solution
A computer program and three-dimensional sensor control device that convert orthogonal coordinates of data points into polar coordinates, locate the farthest data points in each sub-region, and convert them back to orthogonal coordinates to generate environment maps, effectively excluding moving objects as landmarks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If point-cloud matching is performed using conventional SLAM, then environment mapping can be carried out, but the processing load becomes excessively high requiring high-spec computers
Solution Approach 1:
The patent extracts only the essential information needed for environment mapping by selecting the farthest data point in each sub-region rather than performing comprehensive point-cloud matching. This extraction approach obtains sufficient environmental structure information while avoiding the high processing load of full point-cloud algorithms.
2Reliability
If point-cloud matching is performed to generate environment maps, then mapping can be achieved, but the processing time increases significantly
Solution Approach 1:
The patent segments the photographing range into multiple sub-regions and processes each sub-region independently to find the farthest data point. This segmentation divides the complex global matching problem into simpler local problems, significantly reducing processing time while maintaining environment map generation capability.
3Reliability
If SLAM is used to map environments with moving objects, then mapping can be performed, but moving objects are incorrectly recognized as landmarks causing map errors
Solution Approach 1:
The patent performs preliminary action by capturing multiple photograph frames before generating the environment map. By collecting data over time and selecting the farthest points across multiple frames, the system establishes a stable reference of the static environment before any moving objects can interfere with landmark recognition.
Solution Approach 2:
Instead of using all detected data points including those from moving objects, the patent applies partial action by selectively using only the farthest data point in each sub-region. This partial selection inherently filters out moving objects since they typically occupy closer positions, thus improving landmark recognition accuracy without requiring complex object classification.
4Quantity of substance
If cloud of data points from three-dimensional sensor is used, then three-dimensional information can be obtained, but the data density is insufficient compared to still images for adequate matching
Solution Approach 1:
The patent changes the parameter of data selection from using all detected points to selecting only the farthest point in each sub-region. This parameter change transforms the sparse point-cloud data into a sufficient representation by strategically choosing points that define the environmental boundaries, effectively compensating for the lower data density compared to still images.
Data Source
AI summary
A three-dimensional sensor control device performs processes of acquiring orthogonal coordinates of data points found in a plurality of photograph frames photographed by at least one three-dimensional sensor with a fixed photographing range and converting the orthogonal coordinates of the data points into polar coordinates using a first formula for conversion from orthogonal coordinates to polar coordinates The device then performs searching each sub-region of the photographing range to locate a set of polar coordinates representing a farthest data point that is located farthest in distance from the three-dimensional sensor in a respective sub-region and converting the polar coordinates of a collection of the farthest data points back into orthogonal coordinates using a second formula for conversion from the polar coordinates to the orthogonal coordinates, The device outputs environment map data of the photographing range using the orthogonal coordinates of the collection of the farthest data points.