Multi-Task Neural Network for Position Estimation in Large-Scale Indoor Environments

Abstract

Pedestrian localization within large-scale multi-building/multi-floor indoor environments remains a challenging task. Fingerprinting-based approaches are particularly suited for such large-scale deployments due to their low requirements of hardware installments. Recently, the fingerprinting problem has been addressed by deep learning. Existing models are mostly task specific by providing floor classification or position estimation within a small area. A strategy to support localization within large-scale environments is to sequentially apply hierarchical models. This has several drawbacks including missing scalability and increased deployment complexity on smartphones. We propose a unifying approach based on training a single neural network that classifies the building/floor and predicts the position in a single forward-pass of the network. Our model classifies a grid cell and performs within grid cell regression, which solves the performance degradation of applying regression within large areas. To reduce the error in case of misclassified grid cells, we propose a novel technique called multi cell encoding learning (multi-CEL), where a model simultaneously learns several redundant position representations within an overlapping grid cell encoding. For three public WLAN fingerprinting datasets, we demonstrate that multi-CEL surpasses existing state-of-the-art multi-task learning neural networks and even outperforms regression neural networks explicitly trained for 2D-positioning by up to 17%.

HyTra: Hyperclass Transformer for WiFi Fingerprinting-based Indoor Localization

“…However, we obtain the highest floor accuracy when we directly treat the PCA-transformed input as feature embeddings. [30] 99.5% 91.26% Hierarchical RNN [35] 100% 95.23 2D-CNN (m-CEL) [34] -95.30% CNNLoc [23] 100% 96.03% HyTra (1,1) 100% 94.33% HyTra (1,2) 99.91% 88.12% HyTra (2,3) 100% 96.47%…”

“…DeepLocBox achieves a best score of 99.64% and 92.62% accuracy for building and floor classification on the UJI dataset out of 10 trials. In [34], Laska and Blakenbach propose multi-cell encoding learning to solve multi-task learning problems using a single forward pass network. Using a CNN backbone, the proposed network simultaneously classifies grid cells and does in-cell regression to achieve 95.3% accuracy in building and floor classification, also 7.18 meters mean positioning error.…”

GeoAI: a review of artificial intelligence approaches for the interpretation of complex geomatics data

“…We are confident that this review offers rich information and improves the understanding of the research issues related to the use of AI with geomatics data, as well as helping to inform researchers about whether and how AI methods and techniques could help in the creation of applications in various fields. This paper thus paves the way for further research and pinpoints key gaps that serve to provide insights for future improvements, especially considering the complexity introduced by image fusion methods and multi-task learning (S. Laska and Blankenbach, 2022). Future research directions include the improvement of the algorithms to use other comprehensive features, thereby achieving better performance.…”

Fingerprinting-based indoor localization: A Deep Learning Approach

“…Also, various convolutional neural networks are applied in the same context. For instance, a multi-cell encoding learning (m-CEL) technique based on fingerprinting is proposed in [31] for estimating position in substantial indoor environments. Using a single forward pass network, this multi-task learning method (m-CEL) addresses the difficulties of building and floor classification.…”