Denoising UWB Radar Data for Human Activity Recognition Using Convolutional Autoencoders

Lafontaine, Virgile; Bouchard, Kévin; Maître, Julien; Gaboury, Sébastien

doi:10.1109/access.2023.3300224

Cited by 5 publications

(4 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our laboratory, the Laboratoire d'Intelligence Ambiante pour la Reconnaissance d'Activité (LIARA), has been at the forefront of research and development in the domain of smart homes aimed at facilitating aging in place for almost two decades [12,21,23,25,29]. Over the years, our research has focused on creating intelligent environments that seamlessly integrate pervasive sensing technologies with user-friendly interfaces to enhance the quality of life of older adults.…”

Section: The Smart Home Setupmentioning

confidence: 99%

“…For instance, within the LIARA smart While the smart home prototype utilized in this project leverages a multitude of sensing technologies, it is important to note that our laboratory possesses a broader spectrum of advanced tools and resources. For instance, technologies such as the power analyzer, which is typically installed on the main electrical panel, and UWB sensors, offer additional capabilities for monitoring and analysis [25]. Furthermore, our lab is equipped with interactive devices designed to facilitate seamless communication and interaction with users.…”

Section: Data-driven Orientationmentioning

confidence: 99%

“…Very often, smart homes still rely on very simple sensors due to the cost and the complexity of processing the data [23]. With the recent deep learning architectures, researchers are now including more advanced sensors such as passive RFID [24], ultrawideband (UWB) radars [25], and cameras [26]. While deep learning can readily extract features and transform complex data to learn meaningful patterns, in smart homes, for some specific tasks, it does not cover all scenarios and all needs.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

AtomGID: An Atomic Gesture Identifier for Qualitative Spatial Reasoning

Bouchard,

Bouchard

2024

Applied Sciences

Self Cite

View full text Add to dashboard Cite

In this paper, we present a novel non-deep-learning-based approach for real-time object tracking and activity recognition within smart homes, aiming to minimize human intervention and dataset requirements. Our method utilizes discreet, easily concealable sensors and passive RFID technology to track objects in real-time, enabling precise activity recognition without the need for extensive datasets typically associated with deep learning techniques. Central to our approach is AtomGID, an algorithm tailored to extract highly generalizable spatial features from RFID data. Notably, AtomGID’s adaptability extends beyond RFID to other imprecise tracking technologies like Bluetooth beacons and radars. We validate AtomGID through simulation and real-world RFID data collection within a functioning smart home environment. To enhance recognition accuracy, we employ a clustering adaptation of the flocking algorithm, leveraging previously published Activities of Daily Living (ADLs) data. Our classifier achieves a robust classification rate ranging from 85% to 93%, underscoring the efficacy of our approach in accurately identifying activities. By prioritizing non-deep-learning techniques and harnessing the strengths of passive RFID technology, our method offers a pragmatic and scalable solution for activity recognition in smart homes, significantly reducing dataset dependencies and human intervention requirements.

show abstract

Section: The Smart Home Setupmentioning

confidence: 99%

Section: Data-driven Orientationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

AtomGID: An Atomic Gesture Identifier for Qualitative Spatial Reasoning

Bouchard,

Bouchard

2024

Applied Sciences

Self Cite

View full text Add to dashboard Cite

show abstract

“…Many countries have proposed "Aging in Place" to assist older adults to live in their own homes and communities safely, independently, and comfortably, regardless of age, income, or ability level [2][3][4]. Human activity recognition (HAR) based on sensors has made significant contributions to human-centric applications in smart homes for elderly care, encompassing home automation, ambient assisted living (AAL) systems, and remote monitoring [5,6]. Among the three main types of sensors used-cameras, wearable, and ambient sensors-each has its predominant advantages and limitations.…”

Section: Introductionmentioning

confidence: 99%

Enhancing Human Activity Recognition in Smart Homes with Self-Supervised Learning and Self-Attention

Chen,

Gouin-Vallerand,

Bouchard

et al. 2024

Sensors

Self Cite

View full text Add to dashboard Cite

Deep learning models have gained prominence in human activity recognition using ambient sensors, particularly for telemonitoring older adults’ daily activities in real-world scenarios. However, collecting large volumes of annotated sensor data presents a formidable challenge, given the time-consuming and costly nature of traditional manual annotation methods, especially for extensive projects. In response to this challenge, we propose a novel AttCLHAR model rooted in the self-supervised learning framework SimCLR and augmented with a self-attention mechanism. This model is designed for human activity recognition utilizing ambient sensor data, tailored explicitly for scenarios with limited or no annotations. AttCLHAR encompasses unsupervised pre-training and fine-tuning phases, sharing a common encoder module with two convolutional layers and a long short-term memory (LSTM) layer. The output is further connected to a self-attention layer, allowing the model to selectively focus on different input sequence segments. The incorporation of sharpness-aware minimization (SAM) aims to enhance model generalization by penalizing loss sharpness. The pre-training phase focuses on learning representative features from abundant unlabeled data, capturing both spatial and temporal dependencies in the sensor data. It facilitates the extraction of informative features for subsequent fine-tuning tasks. We extensively evaluated the AttCLHAR model using three CASAS smart home datasets (Aruba-1, Aruba-2, and Milan). We compared its performance against the SimCLR framework, SimCLR with SAM, and SimCLR with the self-attention layer. The experimental results demonstrate the superior performance of our approach, especially in semi-supervised and transfer learning scenarios. It outperforms existing models, marking a significant advancement in using self-supervised learning to extract valuable insights from unlabeled ambient sensor data in real-world environments.

show abstract