A telematics device is a vehicle instrument that comes preinstalled by the vehicle manufacturer or can be added later. The device records information about driving behavior, including speed, ...acceleration, and turning force. When connected to vehicle computers, the device can also provide additional information regarding the mechanical usage and condition of the vehicle. All of this information can be transmitted to a central database via mobile networks. The information provided has led to new services such as Usage Based Insurance (UBI). A range of consultants, industry commentators and academics have produced an abundance of projections on how telematics information will allow the introduction of services from personalized insurance, bespoke entertainment and advertise and vehicle energy optimization, particularly for Electric Vehicles (EVs). In this paper we examine these potential services against a backdrop of nascent regulatory limitations and against the technical capacity of the devices. Using a case study approach, we examine three applications that can use telematics information. We find that the expectations of service providers will be significantly tempered by regulatory and technical hurdles. In our discussion we detail these limitations and suggest a more realistic rollout of ancillary services.
Event cameras, unlike traditional frame-based cameras, excel in detecting and reporting changes in light intensity on a per-pixel basis. This unique technology offers numerous advantages, including ...high temporal resolution, low latency, wide dynamic range, and reduced power consumption. These characteristics make event cameras particularly well-suited for sensing applications such as monitoring drivers or human behavior. This paper presents a feasibility study on the using a multitask neural network with event cameras for real-time facial analytics. Our proposed network simultaneously estimates head pose, eye gaze, and facial occlusions. Notably, the network is trained on synthetic event camera data, and its performance is demonstrated and validated using real event data in real-time driving scenarios. To compensate for global head motion, we introduce a novel event integration method capable of handling both short and long-term temporal dependencies. The performance of our facial analytics method is quantitatively evaluated in both controlled lab environments and unconstrained driving scenarios. The results demonstrate that useful accuracy and computational speed is achieved by the proposed method to determining head pose and relative eye-gaze direction. This shows that neuromorphic facial analytics can be realized in real-time and are well-suited for edge/embedded computing deployments. While the improvement ratio in comparison to existing literature may not be as favorable due to the unique event-based vision approach employed, it is crucial to note that our research focuses specifically on event-based vision, which offers distinct advantages over traditional RGB vision. Overall, this study contributes to the emerging field of event-based vision systems and highlights the potential of multitask neural networks combined with event cameras for real-time sensing of human subjects. These techniques can be applied in practical applications such as driver monitoring systems, interactive human-computer systems and for human behavior analysis.
Driver monitoring systems (DMS) are a key component of vehicular safety and essential for the transition from semi-autonomous to fully autonomous driving. Neuromorphic vision systems, based on event ...camera technology, provide advanced sensing in motion analysis tasks. In particular, the behaviours of drivers' eyes have been studied for the detection of drowsiness and distraction. This research explores the potential to extend neuromorphic sensing techniques to analyse the entire facial region, detecting yawning behaviours that give a complimentary indicator of drowsiness. A second proof of concept for the use of event cameras to detect the fastening or unfastening of a seatbelt is also developed. Synthetic training datasets are derived from RGB and Near-Infrared (NIR) video from both private and public datasets using a video-to-event converter and used to train, validate, and test a convolutional neural network (CNN) with a self-attention module and a recurrent head for both yawning and seatbelt tasks. For yawn detection, respective F1-scores of 95.3% and 90.4% were achieved on synthetic events from our test set and the "YawDD" dataset. For seatbelt fastness detection, 100% accuracy was achieved on unseen test sets of both synthetic and real events. These results demonstrate the feasibility to add yawn detection and seatbelt fastness detection components to neuromorphic DMS.
Autonomous vehicles (AV) have advanced considerably over the past decade and their potential to reduce road accidents is without equal. That said, the evolution towards fully automated driving will ...be accompanied by new and unfamiliar risks. The deployment of AVs hinges on the premise that they are considerably safer than human drivers. However, the ability of manufacturers, insurers and regulators to quantifiably demonstrate this risk reduction, relative to humans, presents a major barrier. Based on accident rates, it will likely take hundreds of millions of autonomous miles to derive statistically meaningful results. This paper addresses this issue and proposes a novel means of quantifying AV accident risks by benchmarking against a more familiar and quantifiable risk - Human Behaviour. This method is used to proactively quantify AV safety relative to human drivers. Currently, anomalous driving behaviour stems from human susceptibilities such as fatigue or aggression. We exploit this observation and explore AV driving behaviour where driving anomalies are symptoms of technology errors. The comparative behaviours of AV and safe human driving can be used to measure AV accident risk. An end-to-end model AV is simulated using Convolutional Neural Networks (CNN) to compare human and AV driving behaviours. Using a machine learning technique called Gaussian Processes (GP), contextual driving anomalies are detected, the frequency and severity of which are used to derive a risk score. This paper offers a starting point for addressing the challenges surrounding AV risk modelling.
The proliferation of technologies embedded in connected and autonomous vehicles (CAVs) increases the potential of cyber-attacks. The communication systems between vehicles and infrastructure present ...remote attack access for malicious hackers to exploit system vulnerabilities. Increased connectivity combined with autonomous driving functions pose a considerable threat to the vast socioeconomic benefits promised by CAVs. However, the absence of historical information on cyber-attacks mean that traditional risk assessment methods are rendered ineffective. This paper proposes a proactive CAV cyber-risk classification model which overcomes this issue by incorporating known software vulnerabilities contained within the US National Vulnerability Database into model building and testing phases. This method uses a Bayesian Network (BN) model, premised on the variables and causal relationships derived from the Common Vulnerability Scoring Scheme (CVSS), to represent the probabilistic structure and parameterisation of CAV cyber-risk. The resulting BN model is validated with an out-of-sample test demonstrating nearly 100% prediction accuracy of the quantitative risk score and qualitative risk level. The model is then applied to the use-case of GPS systems of a CAV with and without cryptographic authentication. In the use case, we demonstrate how the model can be used to predict the effect of risk reduction measures.
The transition to semiautonomous driving is set to considerably reduce road accident rates as human error is progressively removed from the driving task. Concurrently, autonomous capabilities will ...transform the transportation risk landscape and significantly disrupt the insurance industry. Semiautonomous vehicle (SAV) risks will begin to alternate between human error and technological susceptibilities. The evolving risk landscape will force a departure from traditional risk assessment approaches that rely on historical data to quantify insurable risks. This article investigates the risk structure of SAVs and employs a telematics‐based anomaly detection model to assess split risk profiles. An unsupervised multivariate Gaussian (MVG) based anomaly detection method is used to identify abnormal driving patterns based on accelerometer and GPS sensors of manually driven vehicles. Parameters are inferred for vehicles equipped with semiautonomous capabilities and the resulting split risk profile is determined. The MVG approach allows for the quantification of vehicle risks by the relative frequency and severity of observed anomalies and a location‐based risk analysis is performed for a more comprehensive assessment. This approach contributes to the challenge of quantifying SAV risks and the methods employed here can be applied to evolving data sources pertinent to SAVs. Utilizing the vast amounts of sensor‐generated data will enable insurers to proactively reassess the collective performances of both the artificial driving agent and human driver.
Event cameras contain emerging, neuromorphic vision sensors that capture local-light intensity changes at each pixel, generating a stream of asynchronous events. This way of acquiring visual ...information constitutes a departure from traditional frame-based cameras and offers several significant advantages — low energy consumption, high temporal resolution, high dynamic range and low latency. Driver monitoring systems (DMS) are in-cabin safety systems designed to sense and understand a drivers physical and cognitive state. Event cameras are particularly suited to DMS due to their inherent advantages. This paper proposes a novel method to simultaneously detect and track faces and eyes for driver monitoring. A unique, fully convolutional recurrent neural network architecture is presented. To train this network, a synthetic event-based dataset is simulated with accurate bounding box annotations, called Neuromorphic-HELEN. Additionally, a method to detect and analyse drivers’ eye blinks is proposed, exploiting the high temporal resolution of event cameras. Behaviour of blinking provides greater insights into a driver level of fatigue or drowsiness. We show that blinks have a unique temporal signature that can be better captured by event cameras.
This article aims to introduce a degree of technological and ethical realism to the framing of autonomous vehicle perception and decisionality. The objective is to move the socioethical dialog ...surrounding autonomous vehicle decisionality from the dominance of "trolley framings" to more pressing ethical issues. The article argues that more realistic ethical framings of autonomous vehicle technologies should focus on the matters of HMI, machine perception, classification, and data privacy, which are some distance from the decisionality framing premise of the MIT Moral Machine experiment. To support this claim the article appeals to state-of-the-art technologies and emerging technologies concerning autonomous vehicle perception and decisionality, as a means to inform and frame ethical contexts. This is further supported by considering a context specific ethical framing for each time phase we anticipate regarding emerging autonomous vehicle technology.
Autonomous vehicles (AVs) are expected to considerably improve road safety. That said, accident risk will continue to inflict societal costs. The ability to manage and measure these risks is ...fundamental to ensure societal acceptance and public adoption of AVs. In particular, the ability to quantitatively compare the safety of AVs relative to human drivers is crucial. Managing risk exposures through driving operational design domains (ODD) will also become prevalent. Ultimately, the deployment of AVs will hinge on the premise that they are safer than humans. In this paper, we posit a methodology to quantitatively evaluate AV risks and minimise their risk exposure once they are publically available. Two contributions are offered. First, we provide a proactive means of evaluating AV risks based on driving behaviour and safety-critical events. This offers statistically meaningful comparisons between humans and AVs given the limitation of current historical data. Second, we propose a novel risk-aware path planning methodology for AVs based on telematics behavioural data. Driving data from a cohort of young human drivers over roughly 270,000 km in Ireland is used to demonstrate the posited methodology. An unsupervised geostatistical tool called Kernel Density Estimation (KDE) is used to identify “behavioural hotspots” and the risk exposure at each edge or road segment is modelled. The results are incorporated into a path planning algorithm to find safe route paths for AVs, minimising risk exposures. In addition, Self-Organising Maps (SOM) are employed to identify similar risk groups and individual spatial risk patterns are considered.
The increasing accessibility of mobility datasets has enabled research in green mobility, road safety, vehicular automation, and transportation planning and optimization. Many stakeholders have ...leveraged vehicular datasets to study conventional driving characteristics and self-driving tasks. Notably, many of these datasets have been made publicly available, fostering collaboration, scientific comparability, and replication. As these datasets encompass several study domains and contain distinctive characteristics, selecting the appropriate dataset to investigate driving aspects might be challenging. To the best of the authors’ knowledge, this is the first paper that performs a systematic review of a substantial number of vehicular datasets covering various automation levels. In total, 103 datasets have been reviewed, 35 of which focused on naturalistic driving, and 68 on self-driving tasks. The paper gives researchers the possibility of analyzing the datasets’ principal characteristics and their study domains. Most naturalistic datasets have been centered on road safety and driver behavior, although transportation planning and eco-driving have also been studied. Furthermore, datasets for autonomous driving have been analyzed according to their target self-driving tasks. A particular focus has been placed on data-driven risk assessment for the vehicular ecosystem. It is observed that there exists a lack of relevant publicly available datasets that challenge the creation of new risk assessment models for semi- and fully automated vehicles. Therefore, this paper conducts a gap analysis to identify possible approaches using existing datasets and, additionally, a set of relevant vehicular data fields that could be incorporated in future data collection campaigns to address the challenge.