Image-Based Indoor Positioning Accuracy Improvement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current indoor positioning systems (IPSs) face limitations in accuracy, often estimating device locations to within several meters, which is insufficient for critical applications, such as emergency situations where precise location differentiation between safe and hazardous areas is necessary.
Innovation Solution
An image-based framework that converts temporal and spatial information into 2-D images, leveraging deep learning to refine location estimates by inputting these images into a machine learning model trained to minimize location estimation errors, specifically using a 3-D convolutional deep neural network to align location estimates with path segments on a floor map.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional indoor positioning systems use RF signal processing to estimate device locations, then the system can provide location estimates, but the accuracy is limited to several meters which is insufficient for critical applications
Solution Approach 1:
The patent converts spatial location data into image representations, transforming the positioning problem from traditional coordinate-based estimation to image-based pattern recognition. This dimensional transformation enables the use of deep learning techniques that can extract complex spatial features, achieving sub-meter accuracy by analyzing visual patterns in the generated images rather than relying solely on RF signal processing
Solution Approach 2:
The patent introduces an image generation module as an intermediary between the RF signal processing system and the location estimation algorithm. This intermediary converts raw spatial information and device trajectories into visual representations that can be processed by machine learning models, bridging the gap between traditional positioning data and advanced AI-based accuracy improvement
2Measurement precision
If the system increases location estimation accuracy using advanced methods, then measurement precision improves, but the computational complexity and processing requirements increase
Solution Approach 1:
The patent performs preliminary actions by pre-generating images from spatial information and device trajectories before location estimation is needed. The image generation module creates visual representations in advance, allowing the deep learning model to process pre-prepared data rather than performing complex computations in real-time, thus reducing operational complexity while maintaining high accuracy
Data Source
AI summary
In one embodiment, a service obtains spatial information regarding a physical area. The service estimates locations of a device within the physical area over time, based on wireless signals sent by the device. The service generates a set of images based on the spatial information regarding the physical area and on the estimated locations of the device within the physical area over time. The service updates an estimated location of the device by inputting the generated set of images to a machine learning model trained to minimize a location estimation error.