Identifying accident patterns is one of the most vital research foci of driving analysis. Environmental or safety applications and the growing area of fleet management all benefit from accident detection contributions by minimizing the risk vehicles and drivers are subject to, improving their service and reducing overhead costs. Some solutions have been proposed in the past literature for automated accident detection that are mainly based on traffic data or external sensors. However, traffic data can be difficult to access, while external sensors can end up being difficult to set up and unreliable, depending on how they are used. Additionally, the scarcity of accident detection data has limited the type of approaches used in the past, leaving in particular, machine learning (ML) relatively unexplored. Thus, in this paper, we propose a ML framework for automated car accident detection based on mutimodal in-car sensors. Our work is a unique and innovative study on detecting real-world driving accidents by applying state-of-the-art feature extraction methods using basic sensors in cars. In total, five different feature extraction approaches, including techniques based on feature engineering and feature learning with deep learning are evaluated on the strategic highway research program (SHRP2) naturalistic driving study (NDS) crash data set. The main observations of this study are as follows: (1) CNN features with a SVM classifier obtain very promising results, outperforming all other tested approaches. (2) Feature engineering and feature learning approaches were finding different best performing features. Therefore, our fusion experiment indicates that these two feature sets can be efficiently combined. (3) Unsupervised feature extraction remarkably achieves a notable performance score.

Driving a car is an activity that became necessary for exploration, even when living in the present world. Research exploring the topic of safety on the roads has therefore become increasingly relevant. In this paper, we propose a recognition algorithm based on physiological signals acquired from JINS MEME ES_R smart glasses (electrooculography, acceleration and angular velocity) to classify four commonly encountered road types: city road, highway, housing estate and undeveloped area. Data from 30 drivers were acquired in real driving conditions. Hand-crafted statistical features were extracted from the physiological signals to train and evaluate a random forest classifier. We achieved an overall accuracy, precision, recall and F1 score of 87.64%, 86.30%, 88.12% and 87.08% on the test dataset, respectively.

Changes in gait are associated with an increased risk of falling and may indicate the presence of movement disorders related to neurological diseases or age-related weakness. Continuous monitoring based on inertial measurement unit (IMU) sensor data can effectively estimate gait parameters that reflect changes in gait dynamics. Monitoring using a waist-level IMU sensor is particularly useful for assessing such data, as it can be conveniently worn as a sensor-integrated belt or observed through a smartphone application. Our work investigates the efficacy of estimating gait events and gait parameters based on data collected from a waist-worn IMU sensor. The results are compared to measurements obtained using a GAITRite^® system as reference. We evaluate two machine learning (ML)-based methods. Both ML methods are structured as sequence to sequence (Seq2Seq). The efficacy of both approaches in accurately determining gait events and parameters is assessed using a dataset comprising 17,643 recorded steps from 69 subjects, who performed a total of 3588 walks, each covering approximately 4 m. Results indicate that the Convolutional Neural Network (CNN)-based algorithm outperforms the long short-term memory (LSTM) method, achieving a detection accuracy of 98.94% for heel strikes (HS) and 98.65% for toe-offs (TO), with a mean error (ME) of 0.09 ± 4.69 cm in estimating step lengths.

Changes in gait are associated with an increased risk of falling and may indicate the presence of movement disorders related to neurological diseases or age-related weakness. Continuous monitoring based on inertial measurement unit (IMU) sensor data can effectively estimate gait parameters that reflect changes in gait dynamics. Monitoring using a waist-level IMU sensor is particularly useful for assessing such data, as it can be conveniently worn as a sensor-integrated belt or through a smartphone application. Our work investigates the efficacy of estimating gait events and gait parameters based on data collected from a waist-worn IMU sensor. Results are compared to measurements obtained using a GAITRite\textsuperscript{\textregistered} system as reference. We evaluate two machine learning (ML) based methods. Both ML methods are structured as sequence-to-sequence models (Seq2Seq). The efficacy of both approaches in accurately determining gait events and parameters is assessed using a dataset consisting of 17,643 recorded steps from 69 subjects, who performed a total of 3,588 walks, each covering approximately 4 meters. Results indicate that the CNN-based algorithm outperforms the LSTM method, achieving a detection accuracy of 98.94% for heel strikes and 98.65% for toe offs, with a mean error of 0.09 ± 4.69 cm in estimating step lengths.