Falls are one of the main health risks among the elderly. A fall detection system based on inertial sensors can automatically detect fall event and alert a caregiver for immediate assistance, so as to reduce injuries causing by falls. Nevertheless, most inertial sensor-based fall detection technologies have focused on the accuracy of detection while neglecting quantization noise caused by inertial sensor. In this paper, an activity model based on tri-axial acceleration and gyroscope is proposed, and the difference between activities of daily living (ADLs) and falls is analyzed. Meanwhile, a Kalman filter is proposed to preprocess the raw data so as to reduce noise. A sliding window and Bayes network classifier are introduced to develop a wearable fall detection system, which is composed of a wearable motion sensor and a smart phone. The experiment shows that the proposed system distinguishes simulated falls from ADLs with a high accuracy of 95.67%, while sensitivity and specificity are 99.0% and 95.0%, respectively. Furthermore, the smart phone can issue an alarm to caregivers so as to provide timely and accurate help for the elderly, as soon as the system detects a fall.

The Bluetooth-based wearable fall detection technology faces problems such as short transmission distance, easily interfered by obstacles and high-power consumption. To address these issues, we developed a sensor board integrated with low power ZigBee and MPU6050, which can sample and cache 3-axial acceleration and angular velocity data in sleep mode; we also designed an interrupt-driven algorithm that can collect and transmit the data to the receiving end (namely server) with low power consumption via ZigBee. Additionally, the received data are normalized according to the range specification and cached into a sliding window by the server. Meanwhile, the cached data are mapped into RGB bitmap, and a Fall Detection Convolutional Neural Network (FD-CNN) is designed and trained using the open dataset to identify falls from activities of daily livings according to bitmap. The experimental results show that the average accuracy of this method is 98.61%, while its average sensitivity and specificity are 98.62% and 99.80% respectively. It takes advantage of the strong networking capacity of ZigBee, and the strong computing power of the server, which is very suitable for fall sensing in elderly community with low power and high accuracy.

The activity model based on 3D acceleration and gyroscope is created in this paper, and the difference between the activities of daily living (ADLs) and falls is analyzed at first. Meanwhile, the kNN algorithm and sliding window are introduced to develop a smart device enabled system for fall detection and alert, which is composed of a wearable motion sensor board and a smart phone. The motion sensor board integrated with triaxial accelerometer, gyroscope, and Bluetooth is attached to a custom vest worn by the elderly to capture the reluctant acceleration and angular velocity of ADLs in real time. The stream data via Bluetooth is then sent to a smart phone, which runs a program based on the kNN algorithm and sliding window to analyze the stream data and detect falls in the background. At last, the experiment shows that the system identifies simulated falls from ADLs with a high accuracy of 97.7%, while sensitivity and specificity are 94% and 99%, respectively. Besides, the smart phone can issue an alarm and notify caregivers to provide timely and accurate help for the elderly, as soon as a fall is detected.

3D object detection with LiDAR and camera fusion has always been a challenge for autonomous driving. This work proposes a deep neural network (namely FuDNN) for LiDAR–camera fusion 3D object detection. Firstly, a 2D backbone is designed to extract features from camera images. Secondly, an attention-based fusion sub-network is designed to fuse the features extracted by the 2D backbone and the features extracted from 3D LiDAR point clouds by PointNet++. Besides, the FuDNN, which uses the RPN and the refinement work of PointRCNN to obtain 3D box predictions, was tested on the public KITTI dataset. Experiments on the KITTI validation set show that the proposed FuDNN achieves AP values of 92.48, 82.90, and 80.51 at easy, moderate, and hard difficulty levels for car detection. The proposed FuDNN improves the performance of LiDAR–camera fusion 3D object detection in the car category of the public KITTI dataset.

A fall detection module is an important component of community-based care for the elderly to reduce their health risk. It requires the accuracy of detections as well as maintains energy saving. In order to meet the above requirements, a sensing module-integrated energy-efficient sensor was developed which can sense and cache the data of human activity in sleep mode, and an interrupt-driven algorithm is proposed to transmit the data to a server integrated with ZigBee. Secondly, a deep neural network for fall detection (FD-DNN) running on the server is carefully designed to detect falls accurately. FD-DNN, which combines the convolutional neural networks (CNN) with long short-term memory (LSTM) algorithms, was tested on both with online and offline datasets. The experimental result shows that it takes advantage of CNN and LSTM, and achieved 99.17% fall detection accuracy, while its specificity and sensitivity are 99.94% and 94.09%, respectively. Meanwhile, it has the characteristics of low power consumption.