Phantom Point Filtering for LiDAR-Based Household Appliance Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Contactless sensors in household appliances like LiDAR sensors often produce phantom points due to bright light or transparent materials, leading to erroneous measurements and inaccurate mapping and positioning.
Innovation Solution
A method and apparatus for creating a map of surroundings by transmitting measurement signals, determining point distributions, and discarding points that do not align with a straight line or exceed a predetermined distance from the average location, using normal distribution transformation (NDT) and Mahalanobis distance to filter out phantom points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If contactless sensors (LiDAR, radar) are used to scan surroundings, then mapping and positioning capability is improved, but phantom points are generated leading to measurement errors
Solution Approach 1:
The patent extracts and removes phantom points from the measurement data obtained by contactless sensors. The processing facility identifies and discards phantom points based on spatial distribution analysis, keeping only valid obstacle points for mapping and positioning, thus eliminating the harmful effect of phantom points while preserving the beneficial measurement capability
Solution Approach 2:
The patent introduces an intermediary processing step between sensor measurement and map creation. The processing facility acts as an intermediary that analyzes spatial distributions, determines whether points are phantom or valid, and filters the data before it is used for mapping, thereby resolving the contradiction between measurement capability and reliability
2Area of stationary object
If all detected points are used for mapping, then complete coverage is achieved, but phantom points contaminate the map data
Solution Approach 1:
The patent applies local quality analysis by examining the spatial distribution characteristics of individual points or groups of points. Each point's validity is determined based on its local spatial relationship with other points, allowing the system to maintain high map accuracy by selectively including only locally consistent points while preserving complete coverage of actual obstacles
3Device complexity
If phantom points are retained in measurement data, then simpler processing is used, but mapping and positioning accuracy deteriorates
Solution Approach 1:
The patent performs preliminary filtering of phantom points before the main mapping and positioning operations. By removing phantom points in advance through spatial distribution analysis, the system ensures that subsequent mapping and positioning algorithms work with clean data, maintaining high precision without requiring complex post-processing corrections
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 mapping and positioning accuracy by effectively removing phantom points, ensuring reliable and precise determination of obstacles and free areas, enhancing the navigation and cleaning efficiency of mobile household appliances.
Implementation Method 1
transmitting a measurement signal into the surroundings of the household appliance and receiving reflections of the measurement signal
Data Source
AI summary
A map of the surroundings for a motorized domestic appliance is created by transmitting signals and allocating reflections to points. A local distribution of the points is formed. Points are discarded if a correspondence of the distribution to a straight line fails to reach a predefined degree. In a second variant, points whose distance from an average location exceeds a predefined value are discarded. The surroundings may be divided into tiles. A normal distribution transformation for the points of a tile specifies an average location and a variance of the points. The variance has two eigenvectors in two dimensions, each with an eigenvalue describing the variance contained in the eigenvector. A ratio between the larger and smaller eigenvalues describes an equal or unequal distribution of the variances. If, however, the quotient is, say, 10 or more, a large number of points of the tile can lie along a straight line.