Abstract:In this paper we propose mmFall - a novel fall detection system, which comprises of (i) the emerging millimeter-wave (mmWave) radar sensor to collect the human body's point cloud along with the body centroid, and (ii) a variational recurrent autoencoder (VRAE) to compute the anomaly level of the body motion based on the acquired point cloud. A fall is claimed to have occurred when the spike in anomaly level and the drop in centroid height occur simultaneously. The mmWave radar sensor provides several advantages, such as privacycompliance and high-sensitivity to motion, over the traditional sensing modalities. However, (i) randomness in radar point cloud data and (ii) difficulties in fall collection/labeling in the traditional supervised fall detection approaches are the two main challenges. To overcome the randomness in radar data, the proposed VRAE uses variational inference, a probabilistic approach rather than the traditional deterministic approach, to infer the posterior probability of the body's latent motion state at each frame, followed by a recurrent neural network (RNN) to learn the temporal features of the motion over multiple frames. Moreover, to circumvent the difficulties in fall data collection/labeling, the VRAE is built upon an autoencoder architecture in a semi-supervised approach, and trained on only normal activities of daily living (ADL) such that in the inference stage the VRAE will generate a spike in the anomaly level once an abnormal motion, such as fall, occurs. During the experiment, we implemented the VRAE along with two other baselines, and tested on the dataset collected in an apartment. The receiver operating characteristic (ROC) curve indicates that our proposed model outperforms the other two baselines, and achieves 98% detection out of 50 falls at the expense of just 2 false alarms.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: With the acceleration of global population aging, fall prevention and detection for the elderly have become extremely crucial. Traditional camera - based fall - detection methods have privacy issues, and collecting real - fall data is very difficult and expensive. Especially for the elderly, repeatedly simulating falls is almost an impossible task. To solve these problems, this paper proposes a new fall - detection system **mmFall**, which uses millimeter - wave (mmWave) radar sensors and Hybrid Variational RNN AutoEncoder (HVRAE). Specifically, this system aims to: 1. **Protect privacy**: Compared with traditional visual sensors (such as cameras), millimeter - wave radar sensors can provide privacy protection while being highly sensitive to motion. 2. **Overcome data - collection difficulties**: By adopting a semi - supervised anomaly - detection method and training the model only on normal activities of daily living (ADL) data, the need for a large amount of labeled fall data is avoided. ### Main challenges - **Randomness of radar point clouds**: The point - cloud data obtained by millimeter - wave radars is random, which poses a challenge to model training. - **Difficulties in fall - data collection and annotation**: Traditional supervised - learning methods rely on a large amount of fall data, but such data is difficult to collect and annotate, especially in the elderly population. ### Solutions - **Hybrid Variational RNN AutoEncoder (HVRAE)**: - Use variational inference to process the radar point - cloud data of each frame and infer the latent distribution of human - motion states. - Utilize recurrent neural networks (RNN) to summarize the temporal features of multiple frames, thereby modeling falls as a sequence of events. - In the inference stage, when an abnormal motion (such as a fall) occurs, HVRAE will generate a peak of the anomaly level. ### Experimental results In the experiment, the author compared HVRAE with two other baseline models and tested it on a dataset collected in an apartment environment. The results show that the proposed model achieved 98% accuracy in 50 fall - detections, with only 2 false positives. ### Summary The main contributions of this paper are: 1. Proposing the first semi - supervised fall - detection method based on 4D millimeter - wave - radar point clouds. 2. Introducing variational inference into the learning of radar - point - cloud distributions to deal with the randomness and complexity of data. Through these innovations, this system not only improves the accuracy of fall - detection but also solves the privacy and data - collection problems existing in traditional methods.

mmFall: Fall Detection using 4D MmWave Radar and a Hybrid Variational RNN AutoEncoder

Real-Time Fall Detection Using Mmwave Radar

Unobtrusive Human Fall Detection System Using mmWave Radar and Data Driven Methods

Fall Detection System Using Millimeter-Wave Radar Based on Neural Network and Information Fusion

A Fall Detection System Based on FMCW Radar Range-Doppler Image and Bi-LSTM Deep Learning

Fall Detection System Based on Millimeter Wave Radar and Machine Learning

Fall Detection With UWB Radars and CNN-LSTM Architecture

An Unobtrusive Fall Detection System Using Ceiling-mounted Ultra-wideband Radar

A Survey on Radar-Based Fall Detection

Fall Detection System Based on Point Cloud Enhancement Model for 24 GHz FMCW Radar

Fall Detection Based on Parallel 2DCNN-CBAM With Radar Multidomain Representations

A Real-time Fall Detection System Using ToF Depth Images.

Doppler Radar Sensor-Based Fall Detection Using a Convolutional Bidirectional Long Short-Term Memory Model

SCL-Fall: Reliable Fall Detection Using Mmwave Radar with Supervised Contrastive Learning

Radar-Based Fall Detection: A Survey

DeepFall -- Non-invasive Fall Detection with Deep Spatio-Temporal Convolutional Autoencoders

Human Fall Detection with Ultra-Wideband Radar and Adaptive Weighted Fusion

Human Fall Detection Based on Machine Learning Using a THz Radar System

Environment-independent mmWave Fall Detection with Interacting Multiple Model

A Millimetre-Wave Radar-Based Fall Detection Method Using Line Kernel Convolutional Neural Network