Indoor Robot Localization Using 3D Voxel Lidar Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional localization methods in indoor spaces face reduced accuracy due to congestion from people and structures, making precise robot location estimation challenging.
Innovation Solution
A method and system utilizing a deep learning network for precise localization, which processes spatial information using lidar data and allocates occupancy information to voxels on spherical coordinates, considering features like people and structures, and defines a new occupancy state for blocked spaces to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional localization methods are used in indoor spaces, then the system is simple to implement, but localization accuracy deteriorates due to congestion from people and structures
Solution Approach 1:
The space is divided into multiple voxels (volume elements) that are processed independently. Each voxel represents a discrete portion of the 3D environment, allowing the system to handle complex indoor spaces by breaking them down into manageable units that can be individually analyzed for occupancy and feature extraction
Solution Approach 2:
The patent transitions from 2D localization to 3D localization by introducing vertical dimension through voxel-based processing. This dimensional expansion allows the system to capture and utilize spatial information in three dimensions, improving localization accuracy in congested indoor environments where objects are distributed throughout the volume rather than just on a plane
2Measurement precision
If lidar-based localization is used, then position recognition capability is improved, but performance deteriorates in congested indoor spaces with moving objects
Solution Approach 1:
The system performs preliminary processing of lidar data by pre-segmenting the environment into voxels and pre-identifying occupancy patterns before actual localization occurs. This preliminary structuring of spatial data enables the system to quickly adapt to changes in congested environments without requiring complete re-processing, thereby maintaining reliability when people and objects move
Solution Approach 2:
The system continuously compares current voxel-based spatial information with stored reference maps, using the differences to refine localization estimates. This feedback mechanism allows the system to adapt to dynamic changes in indoor environments caused by moving people and objects, maintaining reliable localization despite congestion
3Measurement precision
If deep learning network is introduced for feature extraction, then localization accuracy is improved, but processing time and computational complexity increase
Solution Approach 1:
The deep learning network processes segmented voxel data rather than raw point cloud data. By pre-dividing the environment into discrete voxels with defined occupancy properties, the input to the neural network is already structured and reduced in complexity, allowing faster processing while maintaining the ability to extract meaningful features for accurate localization
Solution Approach 2:
The patent transforms raw lidar measurements into voxel occupancy parameters (present/absent, occupied/unoccupied) before feeding data to the deep learning network. This parameter transformation simplifies the input data structure, enabling the network to focus on extracting discriminative features rather than processing raw coordinate data, thereby reducing processing time while improving localization accuracy
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
Enables high-accuracy localization of robots in indoor spaces by comprehensively considering the features of the environment, improving the performance of lidar-based position recognition algorithms.
Implementation Method 1
scanning a surrounding space by using laser outputted from a reference region; processing spatial information about the surrounding space, based on a reflection signal of the laser
Data Source
AI summary
A localization method includes scanning a surrounding space by using laser outputted from a reference region; processing spatial information about the surrounding space based on a reflection signal of the laser; extracting feature vectors to which the spatial information has been reflected by using a deep learning network which uses space vectors including the spatial information as input data; and comparing the feature vectors with preset reference map data, and thus estimating location information about the reference region.