Robot Indoor Positioning With Local Area Deep Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for robots to determine their position in large indoor spaces, such as airports and shopping malls, face challenges due to low GPS precision and high computational requirements, especially when relying on single deep learning networks for position estimation.
Innovation Solution
The method involves dividing a large space into multiple local areas and using specific deep learning networks for each area, with a hierarchical approach that includes a local area classifier and position estimator, utilizing convolutional neural networks and long-short term memory layers to improve accuracy and speed of position estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS is used to acquire position information in large spaces, then the robot can obtain position data, but the precision is low and may fail to perform operations
Solution Approach 1:
The patent divides a large indoor space into multiple local areas, each with its own dedicated deep learning network for position estimation. This segmentation allows the system to achieve high precision positioning within each local area while maintaining reliability across the entire large space, overcoming GPS limitations without requiring a single complex global positioning system.
2Area of stationary object
If a single deep learning network is used for position estimation in large spaces, then the system can provide comprehensive coverage, but the computational requirements and processing time increase significantly
Solution Approach 1:
The patent segments the large space into multiple local areas, each handled by a separate deep learning network. This allows position estimation to be performed in parallel across multiple smaller networks rather than one large network, significantly reducing processing time while maintaining comprehensive coverage through the coordinated operation of all local networks.
Solution Approach 2:
The patent implements local quality by training separate deep learning networks for each local area with characteristics optimized for that specific region. Each network achieves higher estimation accuracy and faster processing for its local area, and the system maintains overall coverage by selecting and using the appropriate local network based on the robot's current area.
3Measurement precision
If deep learning networks are used for position estimation in each local area, then the precision and speed improve, but the device complexity increases
Solution Approach 1:
The patent manages device complexity by segmenting the positioning system into multiple independent but standardized local networks. Each network has the same structural design and can be independently trained and deployed, making the overall system more manageable despite having multiple networks. The complexity is distributed and organized rather than concentrated in a single complex global network.
Data Source
AI summary
A method for estimating a position of a robot in a local area of a large space and a robot and a cloud server that implement such method are provided. The robot includes a local area classifier configured to identify a local area of a plurality of local areas of the space in which the robot moves and a plurality of position estimators configured to estimate the position of the robot in the local area.