No full-text available
To read the full-text of this research,
you can request a copy directly from the author.
Do Automated Vehicles Reduce the Risk of Crashes–Dream or Reality?
Abstract
In the future, the role of the human factor in the driving processes is expected to decrease continuously. At the same time, based on the global trends, the role of computer-supported decision systems and artificial intelligence (AI)-based control solutions increases in relation to driving processes, which carries a significant safety-enhancing potential. To assess the possible social benefits of automated vehicle systems objectively, it is necessary to analyze the possible negative effects in detail as well. Accordingly, the aim of this article is to present a statistical survey of crashes involving automated vehicles today in order to identify and evaluate the factors that are relevant in the crashes. The analyzed data showed that when the autopilot mode was turned off and the human driver made the control decisions, the severity of crashes on straight roads was greater at
=
0.1 significance level than when the vehicle was in autopilot mode and the vehicle system made the control decisions. In addition, if the
significance level is 0.2, then crashes on plain terrain, during the day, or in the speed range of 80-100 km/h are generally less serious for vehicles driven in autopilot mode than for vehicles with autopilot mode turned off. In light of the considerations above, it is also important to emphasize that this paper only investigates crash severity given occurrence but not the probability of occurrence itself.
... Torok et al. [44] applied the Mann-Whitney test to analyze the crash risk involving automated vehicles. ...
... For cyclists, Branion et al. [36] identified the factors associated with higher crash risks, including less frequent cycling, male sex, negative perceptions of cycling in the neighborhood, and residing in areas with high building density. Torok et al. [44] revealed that crashes were more severe when the autopilot mode was turned off in automated vehicles. ...
... Measurement errors in vehicle trajectory data were noted by Xie et al. [64]. Ma et al. [13] discussed the scalability of the data, while studies by Yang et al. [28], Torok et al. [44], and Ko et al. [54] identified the sample size as a major concern. ...
Traffic management systems play a vital role in ensuring safe and efficient transportation on roads. However, the use of advanced technologies in traffic management systems has introduced new safety challenges. Therefore, it is important to ensure the safety of these systems to prevent accidents and minimize their impact on road users. In this survey, we provide a comprehensive review of the literature on safety in traffic management systems. Specifically, we discuss the different safety issues that arise in traffic management systems, the current state of research on safety in these systems, and the techniques and methods proposed to ensure the safety of these systems. We also identify the limitations of the existing research and suggest future research directions.
... Torok et al. [44] applied the Mann-Whitney test to analyze the crash risk involving automated vehicles. ...
... For cyclists, Branion et al. [36] identified the factors associated with higher crash risks, including less frequent cycling, male sex, negative perceptions of cycling in the neighborhood, and residing in areas with high building density. Torok et al. [44] revealed that crashes were more severe when the autopilot mode was turned off in automated vehicles. ...
... Measurement errors in vehicle trajectory data were noted by Xie et al. [64]. Ma et al. [13] discussed the scalability of the data, while studies by Yang et al. [28], Torok et al. [44], and Ko et al. [54] identified the sample size as a major concern. ...
Traffic management systems play a vital role in ensuring safe and efficient transportation on roads. However, the use of advanced technologies in traffic management systems has introduced new safety challenges. Therefore, it is important to ensure the safety of these systems to prevent accidents and minimize their impact on road users. In this survey, we provide a comprehensive review of the literature on safety in traffic management systems. Specifically, we discuss the different safety issues that arise in traffic management systems, the current state of research on safety in these systems, and the techniques and methods proposed to ensure the safety of these systems. We also identify the limitations of the existing research and suggest future research directions.
... While digitalisation can bring many benefits to rail transport, potential threats must be addressed. Some of the main risks of digitalisation for rail transport are outlined below (Török, 2023): ...
Artificial intelligence plays a revolutionary role in modern transport systems. The article discusses the role of artificial intelligence in railway transport and its potential impact on the sector. This article presents different types of artificial intelligence technologies used in this sector, explores the advantages of artificial intelligence in this field, and discusses the challenges associated with using artificial intelligence in rail transport. Artificial Intelligence is revolutionary in rail transport systems by enhancing efficiency, safety, and overall performance. There are several ways in which AI influences rail transport and its impact on the cognitive load of human resources. These factors are examined in this article.
... However, in many countries, the situation is exactly the opposite, with more fatal accidents reported in urban areas than outside of them. The main cause of this is the non-respect of speed limits (Török, 2023). Fig. 5 shows another important statistic, namely that the majority of fatal accidents, which occurred in the European Union in 2016, did not happen due to unfavorable weather conditions -they represented only 20...30% of total accidents (Blincoe et al., 2015). ...
The technology of autonomous driving systems has the potential to bring significant benefits in terms of safety, mobility, and congestion, as well as land use and energy consumption. In this paper, the authors present both the promises and possible dangers of the large-scale implementation of autonomous vehicle technologies in the current road traffic networks. The effects of autonomous mobility technology are complex and require careful consideration. While there is great potential to transform transportation and offer significant improvements in many areas, there are also risks and challenges that need to be addressed to ensure safe, efficient, and equitable deployment of the technology.
... There might be dangerous input/output intervals where we should ignore their application [4] [5]. Beyond AI, for certain highly complex decision-making situations, particularly for vehicular functions based on the interaction between the environment and the vehicle, there are still several problems to be solved to ensure the overall safe operation of the vehicle [6], [7]. Compared to the traditional functional safety approach, the safety of intended functionalities (SOTIF) reflects the challenges of evaluating complex interactions between the system and the environment [8] [9] [10], including also humans [11], [12]. ...
Innovative testing and validation methods are prerequisites concerning Connected, Cooperative, and Automated Mobility (CCAM) as the high number of cooperating participants and concurrent processes critically increase the probability of adverse safety and security incidents. The proposed new approaches deal with this increasing complexity of not currently having generally accepted validation mechanisms. The paper introduces a novel, mathematical model based, scenario identification methodology, facilitating the selection of critical road vehicle traffic scenarios, taking into account different testing objectives, such as maximizing the safety risk of the analyzed system. The presented results verify that applying specific decision models and quantifiable indicators related to the system elements of highly automated mobility systems can significantly contribute to the systematic identification of unsafe corner cases in connected and cooperative autonomous systems.
... While digitalization has the potential to bring many benefits to railway transportation, there are also some potential threats that need to be addressed. Here are some of the key threats of digitalization in rail transport (Török, 2023): ...
Artificial intelligence (AI) has been a revolutionary force in modern transportation systems. In recent years, AI has played an important role in the development of railway transportation. This paper explores the role of AI in railway transportation and the potential impact it may have on the industry. The paper begins by discussing the different types of AI technologies that are being used in railway transportation. It then examines the advantages of AI in railway transportation, including improved safety, increased efficiency, and reduced costs. Finally, the paper discusses the challenges associated with implementing AI in railway transportation and concludes with a discussion of future developments in this area.
Background. Autonomous vehicle (AV) technology has been touted as a means to reduce traffic accidents because computers always pay attention to road conditions and are never intoxicated, which are responsible for most traffic accidents. However, research has been mixed regarding whether AVs actually are involved in fewer collisions than vehicles driven by human operators. Much research has shown that most collisions involving AVs have been collisions in which they have been rear-ended by other vehicles. While this research has suggested that such rear-end collisions are caused by improper maneuvers by the AV or short following distances by the human driver, there has been no research identifying the types of AV behaviors that may result in rear-end collisions. Properly identifying such behaviors would be useful for determining what measures may be most effective in mitigating risks of collisions. Methods. To help identify AV behaviors that may contribute to rear-end collisions, we examined incidents involving AV rideshare vehicles in San Francisco, California in 2023. Descriptions of these incidents were provided in an online database that had been gathered from multiple media sources. Most of these media incident reports were not of collisions, but of incidents that could cause collisions, and therefore could be considered near-miss or potential incidents. Research has shown that evaluating near-miss incidents can provide valuable information for how to reduce the risk of injury incidents. There were 343 separate and verified incidents described in the media. The latter included 18 collision incidents. Results. The results indicated that most of the media-reported incidents (65%) involved AVs that were stopped or stalled in intersections or travel lanes when they had the right of way or exhibited other unexpected or erratic behavior such as sudden lane changes. Such unexpected behavior can result in emergency responses from human drivers, including emergency braking that may result in rear-end collisions. The media reports also included descriptions of a substantial number of incidents (21%) in which the AV committed the types of errors performed by human drivers such as illegal left turns, failing to yield to pedestrians, blocking crosswalks, and running red lights. AV manufacturers claim that AVs will reduce accidents by eliminating the type of human behavior that causes accidents such as inattention and willingly violating traffic laws. These incidents show that AV manufacturers have failed to prevent these human-type behaviors. Discussion. The results are discussed according to basic human factors principles that must be followed to design AVs that may have the best chance of success in truly reducing AV traffic accidents.
Autonomous vehicles and driver assistance systems achieve satisfactory perception of surroundings but lack cognitive capabilities to conduct knowledge-based risk assessment and key focus cognition. Therefore, we propose the initial research that simulates human cognition to decipher traffic scenarios, to direct attention allocation for driving safety. Firstly, our system imitates cognitive processes including sensing, perceiving, reasoning, and judging, as the pipeline to deliver cognitive understanding beyond sensory perception. Secondly, diverse information and knowledge embedded in our criteria, fuzzy systems, and neural networks vitalizes the cognitive processes and thus fosters interpretability and trustworthiness. It goes beyond the visual features and differs from the utilization of condition-action knowledge in trigger rules design. Thirdly, our system achieves a holistic understanding of the entire scene beyond foreground road users or obstacles. It also stresses cognitive-level ‘prominent hazards’ that impact driving safety and deserve priority attentions and provident strategies, beyond visual saliencies or collision-related risks. Experimental comparisons with state-of-the-art models on 22 metrics on two datasets revealed that our system exhibited the overall smallest deviation from ground truths, justifying the effectiveness of our proposed cognition imitation in understanding driving scenarios. Our better results than baseline annotations further verified the capability in assisting situation awareness. The robustness in diverse environmental conditions, alertness to multitype hazards, and the conformity to knowledge indicate the interpretability and trustworthiness in offering forewarnings and directing attentions.
As the problems with managing traffic in cities get worse, this paper looks at a way to make it easier to judge safety and handle risks in the flow of traffic on major roads in cities. By combining artificial driving and smart network connection, the study aims to find better ways to predict, lower, and respond to possible risks on city roads. Getting a lot of data from various sources, such as traffic cameras, sensors, and systems that link vehicles, is part of the research method. Artificial driving models are used to simulate and guess how traffic will move. Cars and buildings can share data in real time through an intelligent network connection. This paper carefully looks at safety by measuring things like collision rates and response times, and then it compares the findings to what happened when traditional traffic management methods were used. Adaptive traffic control and predictive analytics are two types of risk management that are talked about as ways to stop safety problems before they happen. Case studies show examples of how artificial driving and smart network connections have been used successfully to improve road safety in cities. At the end of the paper, issues that came up are discussed, and ideas are given for future research that will help improve how safety is assessed and risks are managed in metro areas where arterial traffic flow is changing.
The New York City (NYC) initiated a new default speed limit law on November 7th, 2014, where speed limits on all road segments without a posted speed limit were reduced from 30 mph to 25 mph. The safety effectiveness of citywide speed limit reduction in an urban setting like NYC has been understudied in the literature. The high-density road network of NYC could lead to a significant spatial spillover effect of speed limit reduction on its neighboring sites. In addition, citywide speed limit reduction exerts much more treatment sites than control sites, which makes it challenging to identify sufficient control sites with similar covariates as the treated ones and thus may lead to confounding bias. Furthermore, there could also exist a time trend in crash observations caused by unobserved factors (e.g., enforcement, driving behaviors). To jointly account for spatial spillover effect, confounding bias, and time trend, this study proposes a novel causal inference approach integrating propensity score matching (PSM) and spatial difference-in-differences (SDID) to estimate the safety effectiveness of citywide speed limit reduction in NYC. The PSM utilizes a logistic generalized additive model (GAM) to capture the nonlinear relationship between covariates and the treatment indicator to reduce bias due to confounding variables. Moreover, the matched data are used to develop the SDID model that simultaneously captures spatial spillover effect and time trend via the extended difference-in-differences (DID) structure. The proposed causal approach suggests that the speed limit reduction would result in a 62.09% decrease in fatal crashes, with the spatial spillover effect found to be statistically significant. However, it does not indicate a significant change in injury and property-damage-only crashes as a result of the speed limit reduction. This study adds to the literature a robust causal inference approach for safety evaluation and provides researchers, practitioners, policy-makers insights into the safety effectiveness of the speed limit reduction in an urban setting.
Objective
We present a model for visual behavior that can simulate the glance pattern observed around driver-initiated, non-critical disengagements of Tesla’s Autopilot (AP) in naturalistic highway driving.
Background
Drivers may become inattentive when using partially-automated driving systems. The safety effects associated with inattention are unknown until we have a quantitative reference on how visual behavior changes with automation.
Methods
The model is based on glance data from 290 human initiated AP disengagement epochs. Glance duration and transition were modelled with Bayesian Generalized Linear Mixed models.
Results
The model replicates the observed glance pattern across drivers. The model’s components show that off-road glances were longer with AP active than without and that their frequency characteristics changed. Driving-related off-road glances were less frequent with AP active than in manual driving, while non-driving related glances to the down/center-stack areas were the most frequent and the longest (22% of the glances exceeded 2 s). Little difference was found in on-road glance duration.
Conclusion
Visual behavior patterns change before and after AP disengagement. Before disengagement, drivers looked less on road and focused more on non-driving related areas compared to after the transition to manual driving. The higher proportion of off-road glances before disengagement to manual driving were not compensated by longer glances ahead.
Application
The model can be used as a reference for safety assessment or to formulate design targets for driver management systems.
Human error is one of the leading causes of accidents. Distraction, fatigue, poor visibility, speeding, and other such errors made by drivers can cause accidents. With the rapid advancements in automation technologies, transportation planners have strived to use Intelligent Transportation Systems (ITS) to minimize human error. In this study, the effect of Autonomous Vehicles (AVs) on the number of potential conflicts at two unsignalized intersections is investigated by using a microsimulation model in PTV Vissim software. For human-driven cars, the factor that is considered for calibration is driver distraction mainly caused by reading or writing text messages on a cellphone while driving. This factor can be estimated using driving simulators. In this paper, five different scenarios were defined for simulation, in addition to the primary state, according to the different market penetration rates of AVs in Vissim. Safety assessment was performed by the Surrogate Safety Assessment Model (SSAM) using Time to Collision (TTC) and Deceleration Rate to Avoid Crashes (DRAC) indicators to determine the number of accidents. It was concluded that the presence of 100% of AVs could reduce the potential for accidents by up to 93%.
In this paper, we aim to identify the significant variables that contribute to the injury severity level of the person in the car when an accident happens and build a statistical model for predicting the maximum injury severity level as well as estimating the potential economic cost in a car accident based on those variables. The General Estimates System data, which is a representative sample of police-reported motor vehicle crashes of all types collected by the National Highway Transportation Safety Administration, from the years 2012 to 2013 is the main data source. Some other data sources such as the car safety rating from the United State Department of Transformation and the state-specific cost of crash deaths fact sheets are also used in the predictive model building process. An interactive system programmed in HyperText Markup Language, Cascading Style Sheets and JavaScript is developed based on the results of predictive modeling. The system is hosted on a website at http://gessmu.azurewebsites.net for public access. The system allows users to input variables that are significant contributors in car accidents and obtain the predicted maximum injury severity level and potential economic cost of a car accident.
Highly Automated Driving technology will be facing major challenges before being pervasively integrated across production vehicles. One of them will be monitoring drivers' state and determining whether they are ready to take over control under certain circumstances. Thus, we have explored their physiological responses and the effects on trust of different scenarios with varying traffic complexity in a driving simulator. Using a mixed repeated measures design, twenty-seven participants were divided in two reliability groups with opposite induced automation reliability expectations -low and high-. We hypothesized that expectations would modulate participants' trust in automation, and consequently, their physiological responses across different scenarios. That is, increasing traffic complexity would also increase participants' arousal, and this would be accentuated or mitigated by automation reliability expectations. Although reliability group differences could not be observed, our results show an increase of physiological activation within high complexity driving conditions (i.e., a mentally demanding non-driving related task and urban scenarios). In addition, we observed a modulation of trust in automation according to the group expectations delivered. These findings provide a background methodology from which further research in driver monitoring systems can benefit and be used to train machine learning methods to classify drivers' state in changing scenarios. This would potentially help mitigate inappropriate take-overs, calibrate trust and increase users' comfort and safety in future Highly Automated Vehicles.
Automation has been changing the types and causes of hazards, and influencing the way in which users interact with complex systems, particularly challenging the notion of human control as a primary basis for hazard mitigation. In this paper, we explore this challenge and use an automotive driving example to examine the distributed nature of the driving task.
We define an Enhanced Vehicle Control Model (VCM) that extends the notion of controllability and joint cognition for highly automated tasks. We apply the model to three contemporary driver assistance systems by undertaking a scenario-based evaluatio. As a conceptual model, the Enhanced VCM shows potential to proactively identify the hazard causes associated with a joint cognitive control. However, to provide utility and to become an effective tool for the system analyst, an accompanying methodology is needed.
Two scenarios have shaped the discussion on the deployment of automated vehicles (AVs). The first is the revolutionary or disruptive scenario, in which a competitor would reach fully automated vehicles (level 5) in one giant leap. The second scenario renders the deployment of AVs as a lengthy process of evolutionary vehicle automation, progressing from specialized, controlled and restricted conditions, i.e. operational design domains (ODDs), to ever more complex ones. But as it becomes more and more clear that there are large uncertainties about how far in the future level 5 AVs become available, in the near future situations in which the driver is replaced completely, i.e. not requiring driver interventions, only seem feasible in specific ODDs and cities will likely be faced by highly automated vehicles (level 4) equipped with automated driving systems (ADSs) for specific ODDs over a longer period. However, cities comprise different street spaces with various conditions which lead to different requirements for the ADSs and could affect the deployment of level 4 AVs and thus the impacts of AVs in cities in the near future. This paper presents a framework and related indicators for assessing the automated drivability, i.e. the suitability of street spaces for the functioning operation of ADSs from a technical-infrastructural point of view, based on the relationship between the current technological development state of ADSs and different (urban) street spaces. Using the case study of the city of Vienna in Austria to apply the framework, a set of indicators for the different components of the framework is calculated and integrated to build an automated drivability index (ADX), illustrating the automated drivability within the street network. The results are a first step to indicate which areas are more suitable for level 4 AVs from a technological point of view, as well as areas where a deployment of AVs would only be possible after larger adjustments of the infrastructure or at considerably lower than intended speeds; thus, detailing the complexity of driving urban street spaces in a yet unprecedented level. Besides other aspects this can give hints for policy makers and other stakeholders where to facilitate the deployment of AVs.
Under proofread
The traditional automotive homologation processes aim to ensure the safety of vehicles on public roads. Autonomous vehicles (AV) with artificial intelligence (AI) are difficult to account for in these conventional processes. This research aims to map and attempt to close the gaps in the areas of testing and approval of such automated and connected vehicles. During our research into the homologation process of traditional vehicles; functional safety issues, challenges of artificial intelligence in safety critical systems, along with questions of cyber security were investigated. Our process focuses on the integration of the already existing functions and prototypes into new products safely. As a key result, we managed to identify the main gaps between information and communication technology and automotive technology: the rigidity of the automotive homologation process, functional safety, artificial intelligence in safety critical areas and we propose a solution.
Autonomous vehicles have the potential to dramatically reduce traffic accidents. This assumption is based on the fact that autonomous vehicles eliminate the impact of a human factor on the occurrence of a traffic accident. Autonomous vehicle’ testing in real traffic conditions is carried out worldwide. In this paper, we analyzed traffic accidents with autonomous vehicles that occurred in the US state of California in the period from 2015 to 2017. In order to better recognize the characteristics of traffic accidents with autonomous vehicles, we were carried out a comparative analysis of traffic accidents with only conventional vehicles at locations where occurred traffic accidents with autonomous vehicles. During the analysis of traffic accidents, we have put emphasis on the type of collision, manoeuvres and errors of the drivers of conventional vehicles that led to the traffic accident. Applying statistical analysis, we were found that the type of collision “rear-end” more often in traffic accidents with autonomous vehicles. Types of collisions “pedestrian” and “broadside” were less in traffic accidents with autonomous vehicles. Drivers’ manoeuvres of conventional vehicles do not differ depending on whether an autonomous vehicle is involved in the traffic accident. Drivers’ errors of conventional vehicles that are more often in accidents with autonomous vehicles are “unsafe speed” and “following too closely”. The obtained results were used to propose measures that will improve communication between autonomous vehicles and drivers’ conventional vehicle.
Information-aware connected and automated vehicles (CAVs) have drawn great attention in recent years due to their potentially significant positive impacts on roadway safety and operational efficiency. In this paper, we conduct an in-depth review of three basic and key interrelated aspects of a CAV: sensing and communication technologies; human factors; and information-aware controller design. First, the different vehicular sensing and communication technologies and their protocol stacks, to provide reliable information to the information-aware CAV controller, are thoroughly discussed. Diverse human factors, such as user comfort, preferences, and reliability, to design the CAV systems for mass adaptation are also discussed. Then, the different layers of a CAV controller (route planning, driving mode execution, and driving model selection) considering human factors and information through connectivity are reviewed. In addition, the critical challenges for the sensing and communication technologies, human factors, and information-aware controller are identified to support the design of a safe and efficient CAV system while considering user acceptance and comfort. Finally, the promising future research directions of these three aspects are discussed to overcome existing challenges to realize a safe and operationally efficient CAV.
The introduction of semi-automated driving systems is expected to mitigate the safety consequences of human error. Observational findings suggest that relinquishing control of vehicle operational control to assistance systems might diminish driver engagement in the driving task, by reducing levels of arousal. In this study, drivers drove a Tesla Model S with Autopilot in both semi-automated and manual modes. Driver behavior was monitored using a combination of physiological and behavioral measures. Compared to manual driving, a reduction in driver physiological activation was observed during semi-automated driving. Also, performance to the peripheral detection task suffered in semi-automated mode, with slower response times recorded in this condition than during manual driving. Taken together, our data suggest that semi-automated driving might not ease safety consequences of human error. Instead, these findings suggest it might cause a drop in driver monitoring, possibly followed by a spike in automation-generated distraction.
In this study, a high accident risk prediction model is developed to analyze traffic accident data, and identify priority intersections for improvement. A database of the traffic accidents was organized and analyzed, and an intersection accident risk prediction model based on different mechanical learning methods was created to estimate the possible high accident risk locations for traffic management departments to use in planning countermeasures to reduce accident risk. Using Bayes’ theorem to identify environmental variables at intersections that affect accident risk levels, this study found that road width, speed limit and roadside markings are the significant risk factors for traffic accidents. Meanwhile, Naïve Bayes, Decision tree C4.5, Bayesian Network, Multilayer perceptron (MLP), Deep Neural Networks (DNN), Deep Belief Network (DBN) and Convolutional Neural Network (CNN) were used to develop an accident risk prediction model. This model can also identify the key factors that affect the occurrence of high-risk intersections, and provide traffic management departments with a better basis for decision-making for intersection improvement. Using the same environmental characteristics as high-risk intersections for model inputs to estimate the degree of risk that may occur in the future, which can be used to prevent traffic accidents in the future. Moreover, it also can be used as a reference for future intersection design and environmental improvements.
In the Vehicular Ad hoc Networks (VANET) environment, recognizing traffic accident events in the driving videos captured by vehicle-mounted cameras is an essential task. Generally, traffic accidents have a short duration in driving videos, and the backgrounds of driving videos are dynamic and complex. These make traffic accident detection quite challenging. To effectively and efficiently detect accidents from the driving videos, we propose an accident detection approach based on spatio–temporal feature encoding with a multilayer neural network. Specifically, the multilayer neural network is used to encode the temporal features of video for clustering the video frames. From the obtained frame clusters, we detect the border frames as the potential accident frames. Then, we capture and encode the spatial relationships of the objects detected from these potential accident frames to confirm whether these frames are accident frames. The extensive experiments demonstrate that the proposed approach achieves promising detection accuracy and efficiency for traffic accident detection, and meets the real-time detection requirement in the VANET environment.
The development of Shared Automated Vehicles (SAVs) is well underway to provide mobility as a service (MaaS) and bring benefits such as reduced traffic congestion, reduced reliance on privately owned vehicles and increased independence to non-drivers. To realise the benefits of SAVs, adoption by private vehicle users is crucial. Previous research has shown this subset of users as the least likely to adopt SAVs, and it is not well understood what factors are important to achieve such adoption. The purpose of this study is to obtain an in-depth understanding of attitudes, perceptions and preferences that influence the acceptance of future SAVs for drivers. This paper presents the results from an online asynchronous focus group study with 21 British drivers as participants. From the analysis, Service Quality, Trust and Price Value emerged as the three most prominent factors to understand user acceptance of SAVs. These three main factors may be of prime importance for convincing naïve private car owners to accept high-speed SAVs. Productive use of travel time has been frequently mentioned in previous research as a benefit of vehicle automation but was scarcely mentioned by participants in this study. Shared Space Quality in introduced as an indicator for Service Quality, together with Security and Trusting Co-passengers as two indicators of Trust. Based on the findings, this paper concludes with a conceptual SAV technology acceptance model is introduced, with the results added as extended model predictors to the Unified Theory of Acceptance and Use of Technology (UTAUT2).
Collision warning systems can improve traffic safety, while their safety benefit may be lessened due to improper risk compensation or system misuse. There are limited studies of advanced safety systems increasing unexpected risky driving behavior, especially with adolescent drivers. This study is designed to address this research gap in two main areas: 1) it seeks to examine whether and how the introduction of advanced driver-assistance systems influences drivers’ risk compensation behavior (e.g., increase of hard braking frequency), and 2) it investigates key factors (e.g., distraction) that contribute to changes in hard braking frequency during driving for both teen and adult drivers. Naturalistic driving data from two previous studies were analyzed in this study with two methods: a hierarchical logistic regression model was used to evaluate the effects of an integrated collision warning system on hard braking behavior, while a Random forests algorithm was applied to model hard braking behavior and to rank the contributing factors by calculating the importance scores. No statistical evidence was observed that the integrated collision warning system significantly changed the likelihood of hard braking for teen or adult drivers. Other factors like distraction, especially visual-manual distraction, had the largest impact on the hard braking behavior, followed by speeding and roadway segments (i.e., at intersections or not). Short time-headways and driving in high-density traffic significantly increased the likelihood of hard braking. Furthermore, the rate of hard braking behavior on surface roads was much higher than on highways, as expected. Compared with straight road segments, hard braking behavior was less likely to occur on curve roads. This study applied an analytical strategy by using both machine learning and statistical analysis methods to achieve high model accuracy and facilitate inference concerning the relationships among variables. Findings in this study can help to improve the design of integrated collision warning systems and the use of autonomous braking systems, and to apply appropriate analysis methods in understanding teen drivers’ behavior changes with those safety systems.
Partially automated vehicles present a large range of information to the driver in order to keep them in-the-loop and engaged with monitoring the vehicle's actions. However, existing research shows that this causes cognitive overload and disengagement from the monitoring task. Adaptive Human Machine Interfaces (HMIs) are an emerging technology that might address this problem, by prioritising the information presented. To date, research aiming to define the driver's glance fixation behaviour in a partially automated vehicle to contribute towards an adaptive interface is scarce. This study used a unique three-day longitudinal driving simulator study design to explore which information drivers in a partially automated vehicle require. Twenty-seven participants experienced nine partially automated driving simulations over three consecutive days. Nine information types, developed from standards, previous studies and industry collaboration, were displayed as discrete icons and presented on a surrogate in-vehicle display. Unique to the literature, this study showed that the recorded eye-tracking data demonstrated that usage of the information types changed with longitudinal driving simulator use. This study provides three key contributions: first, the longitudinal study design suggest that single exposure HMI evaluations may be limited in their assessment. Secondly, this study has methodologically shortlisted a list of nine information types that can be used in future studies to represent future partially automated vehicle interfaces. Finally, this is one of the first studies to characterise glance behaviour for partially automated vehicles. With this knowledge, this study contributes important design recommendations for the development of adaptive interfaces.
There have been a number of high-profile collisions involving Automated Vehicles on the road. Although car manufacturers are making considerable investments into the development of Automated Vehicles, these collisions may deter the public from purchasing and using them. Therefore, solutions need to be developed to prevent these collisions from occurring in the future. One such solution is driver training. A previous literature review identified nine themes which are essential in Automated Vehicle driver training. In this article, a deductive thematic analysis was conducted on five high-profile Automated Vehicle collisions in order to demonstrate the relevance of these themes and to gain insights into how the driver’s behaviour contributed to each collision, thus understand the potential role of training in reducing collisions of this nature. By creating interconnection models for each collision, a consistent pattern emerged. A link was made with the drivers’ attitudes, the accuracy of their mental models and their level of trust in the automation. The automation caused the drivers to become underloaded, which impaired their ability to effectively monitor the automation and the road environment. This could have impaired their situation awareness and their ability to identify and avoid hazards in the path of their vehicle. This analysis suggests that future Automated Vehicle driver training programmes should be multifaceted and cover all nine themes. This analysis has validated these nine driver training themes, so these themes and interconnections can help in the development of a comprehensive training programme for drivers of Automated Vehicles in the future.
Road users and the general population by and large recognise the value of vehicles with automated driving systems and features (otherwise typically known as Autonomous Vehicles (AVs)) in terms of road safety, reduced emissions and convenience, but are still wary of their capability, preferring the ‘comfort zone’ of human operator intervention. Motorcyclists and cyclists conversely, are vulnerable to human fallibility in driving, with the majority of crashes occurring as a consequence of other drivers’ inattention. The transition period associated with the introduction of AVs will require AVs and motorcyclists/cyclists sharing the road for a number of years yet, so we need to understand motorcyclists’/cyclists’ perception of AVs. The question of interest here is whether motorcyclists/cyclists reflect the historical literature in this area by having higher levels of trust for human drivers over AVs, or whether they have higher levels of trust in AVs because it removes the ‘human element’ that has been proven to be particularly dangerous for them. Here we surveyed motorcyclists and cyclists about their trust in human drivers and AVs, and developed a novel suite of questions designed to interrogate the difference between trust in general versus trust as a concept of their own personal safety. Some of the salient outcomes suggest that motorcyclists have medium to low levels of trust for both human drivers and AVs, but are significantly more likely to believe that AVs are safer in terms of their own personal safety, such as prioritising or detecting the rider, compared to human drivers. This relationship varies with age and crash experience. The results here are consistent with the logic that motorcyclists/cyclists have a heightened sense of vulnerability on the road and welcome the introduction of AVs as a way of mitigating personal risk when riding. This insight will be crucial to the subsequent roll-out of AVs in the future.
This paper investigates public perceptions towards potential safety benefits, and safety- and security-related concerns from the future use of autonomous vehicles by utilizing data collected from an online survey. The survey includes responses from 584 individuals from the United States, who responded to a varying range of questions related to autonomous vehicles and their usage. The subsequent exploratory statistical analysis is conducted by employing a novel method, namely the grouped random parameters bivariate probit model with heterogeneity in means. The proposed method accounts for the challenges stemming from the presence of multiple layers of unobserved heterogeneity in the data, and simultaneously offers more insightful results. From the analysis, several socio-demographic characteristics, and driving attitude related characteristics and opinions were found to affect the perceptions towards the safety and security related aspects of autonomous vehicles. The heterogeneity in means approach revealed distinct individual-specific characteristics that affect the peak of the distribution of the parameter density function of the random parameters, adding further clarity to the understanding of the factors affecting individuals’ perceptions towards autonomous vehicles. The findings from this study suggest the ongoing evaluation of public perceptions, and reinforce the requirement of analyzing temporal variations in public perceptions. This can, in turn, aid regulatory and governance entities and autonomous vehicle manufacturers to adapt their strategies and implementation plans accordingly.
While the interest of the transport research community and automotive industry is increasingly turning towards developments and improvements in the field of autonomous vehicles, there is a need for a better understanding of the end users’ preferences regarding perceived passenger comfort, in order to improve acceptance and intention to use. The present study is based on a driving simulator experiment conducted at the University of Leeds Driving Simulator and approaches the issue of comfort via observed speed choice behaviour. Participants drove a series of driving simulator scenarios composed of road segments of different road type, road geometry, risk level at the road edge, and oncoming traffic. They also completed a series of self-report questionnaires, including Arnett’s Inventory of Sensation-seeking. A set of models was developed in order to investigate the effects of road environment and sensation-seeking on speed behaviour. The initial model only considered explanatory variables related to the road environment and accounted for individual unobserved heterogeneity. Past behaviour, serial correlation and heterogeneity in road environment were then introduced in the model specification. The autoregressive disturbance term that accounted for serial correlation was also applied in the form of a random variable and significantly improved model fit. Finally, sensation-seeking was incorporated in the model as a latent variable. The results showed a significant impact of most of the road elements as road type, curvature, risk type at the road edge on observed behaviour, implying a future need for the development of autonomous vehicle controllers that adapt their performance based on the road environment. Moreover, sensation-seeking had a significant and positive effect on speed, which indicates a potential future demand for personalised controllers to meet the users’ individual preferences.
Highly automated vehicles (AV) are in the early stages of deployment and are likely to have significant impacts on the United States transportation system. In particular, a broad deployment of shared, on-demand AVs might significantly impact vehicle ownership and transportation energy consumption; projecting these impacts is essential for climate, infrastructure, and policy planning. However, it seems increasingly likely that AVs will be deployed gradually over a period of decades, in which case there may be geographic or functional variation in their availability. This might occur for a combination of technological, policy, and economic reasons.
This manuscript seeks to advance a new framework for projecting AV impacts, with a particular focus on energy consumption impacts. Specifically, we introduce a framework for AV impacts that allows for AVs catering to specific operating environments or ride types. As a demonstration of this framework, we use the 2009 National Household Transportation Survey (NHTS) to segment US household travel demand based on built environment and ride length. Our framework allows us to specify AV “availability” for each population segment and ride type and use that information to predict the impact of AVs. We analyze a case scenario where shared, on-demand AVs are mostly suited for short trips in highly urbanized environments. We project the impact on household relocation, private vehicle ownership, induced travel demand, and fuel consumption. Utilization of this framework would help identify policy levers for sustainable deployment of AVs.
Introduction:
This study aimed to investigate the characteristics and patterns of the connected and autonomous vehicle (CAV) involved crashes.
Method:
The crash data were collected from the reports of CAV involved crash submitted to the California Department of Motor Vehicles. The descriptive statistics analysis was employed to investigate the characteristics of CAV involved crashes in terms of crash location, weather conditions, driving mode, vehicle movement before crash occurrence, vehicle speed, collision type, crash severity, and vehicle damage locations. The bootstrap based binary logistic regressions were then developed to investigate the factors contributing to the collision type and severity of CAV involved crashes.
Results:
The results suggested that the CAV driving mode, collision location, roadside parking, rear-end collision, and one-way road are the main factors contributing to the severity level of CAV involved crashes. The CAV driving mode, CAV stopped or not, CAV turning or not, normal vehicle turning or not, and normal vehicle overtaking or not are the factors affecting the collision type of CAV involved crashes.
Developing conditionally automated driving systems is on the rise. Vehicles with full longitudinal and latitudinal control will allow drivers to engage in secondary tasks without monitoring the roadway, but users may be required to resume vehicle control to handle critical hazards. The loss of driver’s situational awareness increases the potential for accidents. Thus, the automated systems need to estimate the driver’s ability to resume control of the driving task.
The aim of this study was to assess the physiological behaviour (heart rate and pupil diameter) of drivers. The assessment was performed during two naturalistic secondary tasks. The tasks were the email and the twenty questions task in addition to a control group that did not perform any tasks. The study aimed at finding possible correlations between the driver’s physiological data and their responses to a takeover request. A driving simulator study was used to collect data from a total of 33 participants in a repeated measures design to examine the physiological changes during driving and to measure their takeover quality and response time.
Secondary tasks induced changes on physiological measures and a small influence on response time. However, there was a strong observed correlation between the physiological measures and response time. Takeover quality in this study was assessed using two new performance measures called PerSpeed and PerAngle. They are identified as the mean percentage change of vehicle’s speed and heading angle starting from a take-over request time. Using linear mixed models, there was a strong interaction between task, heart rate and pupil diameter and PerSpeed, PerAngle and response time. This, in turn, provided a measurable understanding of a driver’s future responses to the automated system based on the driver’s physiological changes to allow better decision making. The present findings of this study emphasised the possibility of building a driver mental state model and prediction system to determine the quality of the driver's responses in a highly automated vehicle. Such results will reduce accidents and enhance the driver’s experience in highly automated vehicles.
Road traffic safety is one of the major challenges for the future of smart cities and transportation networks. Despite several solutions exist to reduce the number of fatalities and severe accidents happening daily in our roads, this reduction is smaller than expected and new methods and intelligent systems are needed. The emergency Call is an initiative of the European Commission aimed at providing rapid assistance to motorists thanks to the implementation of a unique emergency number. In this work, we study the problem of classifying the severity of accidents involving Powered Two Wheelers, by exploiting machine learning systems based on features that could be reasonably collected at the moment of the accident. An extended study on the set of features allows to identify the most important factors that enable to distinguish accident severity. The system we develop achieves around 90% of precision and recall on a large, publicly available corpus, using only a set of eleven features.
This article considers how liability questions will be resolved under current Australian laws for automated vehicle (‘AV’) accidents. In terms of the parties that are likely to be held responsible, I argue that whether the human driver remains liable depends on the degree to which the relevant AV is automated, and the degree of control the human driver had over the events leading up to the particular accident. Assuming therefore that human drivers would not be held liable for the majority of highly and fully automated vehicle accidents, plaintiffs will have to establish liability on part of those who manufacture, maintain or contribute to the operation of AVs, under the claims available in Australia's product liability regime.
This article then turns to the problems of proof that plaintiffs are likely to face in establishing AV manufacturer liability in negligence, or in a defective goods claim under Part 3–5 of the Australian Consumer Law (‘ACL’). Firstly, it may be difficult to determine the cause of the AV accident, due to the technical complexity of AVs and due to ongoing concerns as to the explainability of AI-decision making. Secondly, plaintiffs may struggle to prove fault in a negligence claim, or that the vehicle was defective for the purposes of Part 3–5 of the ACL. Essentially, under both actions, manufacturers will be held to a duty to undertake reasonable testing of their AVs. Given that it is currently impracticable to completely test for, and eliminate all AV errors, and due to the broader social utility the technology is likely to offer, plaintiffs may face evidentiary challenges in proving that the manufacturer's testing was unreasonable.
Autonomous Vehicle (AV) technology is quickly becoming a reality on US roads. Testing on public roads is currently undergoing, with many AV makers located and testing in Silicon Valley, California. The California Department of Motor Vehicles (CA DMV) currently mandates that any vehicle tested on California public roads be retrofitted to account for a back-up human driver, and that data related to disengagements of the AV technology be publicly available. Disengagements data is analyzed in this work, given the safety-critical role of AV disengagements, which require the control of the vehicle to be handed back to the human driver in a safe and timely manner. This study provides a comprehensive overview of the fragmented data obtained from AV manufacturers testing on California public roads from 2014 to 2017. Trends of disengagement reporting, associated frequencies, average mileage driven before failure, and an analysis of triggers and contributory factors are here presented. The analysis of the disengagements data also highlights several shortcomings of the current regulations. The results presented thus constitute an important starting point for improvements on the current drafts of the testing and deployment regulations for autonomous vehicles on public roads.
Background: Automated driving represents both challenges and opportunities in highway safety. Google
has been developing self-driving cars and testing them under employee supervision on public roads since
2009. These vehicles have been involved in several crashes, but it is of interest how this testing program
compares to human drivers in terms of safety.
Methods: Google car crashes were coded by type and severity based on narratives released by Google.
Crash rates per million vehicle miles traveled (VMT) were computed for crashes deemed severe enough
to be reportable to police. These were compared with police-reported crash rates for human drivers.
Crash types also were compared.
Results: Google cars had a much lower rate of police-reportable crashes per million VMT than human
drivers in Mountain View, Calif., during 2009-15 (2.19 vs 6.06), but the difference was not statistically
significant. The most common type of collision involving Google cars was when they got rear-ended by
another (human-driven) vehicle. The Google car shared responsibility for only one crash.
Conclusions: These results suggest the Google self-driving car program, while a test program, is safer
than conventional human-driven passenger vehicles; however, currently there is insufficient information
to fully examine the extent to which disengagements affected these results.
Artificial intelligence is everywhere. But before scientists trust it, they first need to understand how machines learn.
Self-driving cars are gaining momentum despite the number of considerable technical and human factor issues that remain controversial. Human acceptance and trust of an automated vehicle to transport people in traffic environments under different driving conditions is a challenging task. Perceived, as well as actual safety will play a major role in accepting automated vehicles. Perceptions however vary from one individual to another due to different reaction times, speed perception, and time constants during dynamical changes etc. The closer the automated vehicle dynamics are with those of a manually driven vehicle the more likely that the comfort level of the automated vehicle user will improve. In this paper we review these issues and discuss how the autopilot personalization feature can help to improve both the perceived and actual driving safety and comfort. We present methodology that allows automatic autopilot personalization based on driver performance models. The methodology takes into account driver's preferences for a particular trip and manual parameters fine-tuning by the driver. We demonstrate how the methodology can be applied on an example of adaptive cruise control and automatic lane change personalization. We support the example with data collected on an experimental vehicle.
Analysis of variance (ANOVA) is a core technique for analysing data in the Life Sciences. This reference book bridges the gap between statistical theory and practical data analysis by presenting a comprehensive set of tables for all standard models of analysis of variance and covariance with up to three treatment factors. The book will serve as a tool to help post-graduates and professionals define their hypotheses, design appropriate experiments, translate them into a statistical model, validate the output from statistics packages and verify results. The systematic layout makes it easy for readers to identify which types of model best fit the themes they are investigating, and to evaluate the strengths and weaknesses of alternative experimental designs. In addition, a concise introduction to the principles of analysis of variance and covariance is provided, alongside worked examples illustrating issues and decisions faced by analysts.
Methods for constructing simultaneous confidence intervals for all possible linear contrasts among several means of normally distributed variables have been given by Scheffé and Tukey. In this paper the possibility is considered of picking in advance a number (say m) of linear contrasts among k means, and then estimating these m linear contrasts by confidence intervals based on a Student t statistic, in such a way that the overall confidence level for the m intervals is greater than or equal to a preassigned value. It is found that for some values of k, and for m not too large, intervals obtained in this way are shorter than those using the F distribution or the Studentized range. When this is so, the experimenter may be willing to select the linear combinations in advance which he wishes to estimate in order to have m shorter intervals instead of an infinite number of longer intervals.
Given C samples, with ni observations in the ith sample, a test of the hypothesis that the samples are from the same population may be made by ranking the observations from from 1 to Σni (giving each observation in a group of ties the mean of the ranks tied for), finding the C sums of ranks, and computing a statistic H. Under the stated hypothesis, H is distributed approximately as χ(C – 1), unless the samples are too small, in which case special approximations or exact tables are provided. One of the most important applications of the test is in detecting differences among the population means.** Based in part on research supported by the Office of Naval Research at the Statistical Research Center, University of Chicago.
The identification of crash hotspots is the first step of the highway safety management process. Errors in hotspot identification may result in the inefficient use of resources for safety improvements and may reduce the global effectiveness of the safety management process. Despite the importance of using effective hotspot identification (HSID) methods, only a few researchers have compared the performance of various methods. In this research, seven commonly applied HSID methods were compared against four robust and informative quantitative evaluation criteria. The following HSID methods were compared: crash frequency (CF), equivalent property damage only (EPDO) crash frequency, crash rate (CR), proportion method (P), empirical Bayes estimate of total-crash frequency (EB), empirical Bayes estimate of severe-crash frequency (EBs), and potential for improvement (PFI). The HSID methods were compared using the site consistency test, the method consistency test, the total rank differences test, and the total score test. These tests evaluate each HSID method's performance in a variety of areas, such as efficiency in identifying sites that show consistently poor safety performance, reliability in identifying the same hotspots in subsequent time periods, and ranking consistency. To evaluate the HSID methods, five years of crash data from the Italian motorway A16 were used.
Let x and y be two random variables with continuous cumulative distribution functions f and g. A statistic U depending on the relative ranks of the x's and y's is proposed for testing the hypothesis . Wilcoxon proposed an equivalent test in the Biometrics Bulletin, December, 1945, but gave only a few points of the distribution of his statistic. Under the hypothesis the probability of obtaining a given U in a sample of and is the solution of a certain recurrence relation involving n and m. Using this recurrence relation tables have been computed giving the probability of U for samples up to . At this point the distribution is almost normal. From the recurrence relation explicit expressions for the mean, variance, and fourth moment are obtained. The 2rth moment is shown to have a certain form which enabled us to prove that the limit distribution is normal if go to infinity in any arbitrary manner. The test is shown to be consistent with respect to the class of alternatives for every x.
Introduction:
In this paper a sensitivity analysis is performed to investigate how big the impact would be on the current ranking of crash locations in Flanders (Belgium) when only taking into account the most serious injury per crash instead of all the injured occupants.
Results:
Results show that this would lead to a different selection of 23.8% of the 800 sites that are currently considered as dangerous.
Conclusions:
Considering this impact quantity, the researchers want to sensitize government that giving weight to the severity of the crash can correct for the bias that occurs when the number of occupants of the vehicles are subject to coincidence. Additionally, probability plots are generated to provide policy makers with a scientific instrument with intuitive appeal to select dangerous road locations on a statistically sound basis. Impact on industry Considering the impact quantity of giving weight to the severity of the crash instead of to all the injured occupants of the vehicle on the ranking of crash sites, the authors want to sensitize government to carefully choose the criteria for ranking and selecting crash locations in order to achieve an enduring and successful traffic safety policy. Indeed, giving weight to the severity of the crash can correct for the bias that occurs when the number of occupants of the vehicles are subject to coincidence. However, it is up to the government to decide which priorities should be stressed in the traffic safety policy. Then, the appropriate weighting value combination can be chosen to rank and select the most dangerous crash locations. Additionally, the probability plots proposed in this paper can provide policy makers with a scientific instrument with intuitive appeal to select dangerous road locations on a statistically sound basis. Note that, in practice, one should not only rank the crash locations based on the benefits that can be achieved from tackling these locations. Future research is also needed to incorporate the costs of infrastructure measures and other actions that these crash sites require in order to enhance the safety on these locations. By balancing these costs and benefits against each other, the crash locations can then be ranked according to the order in which they should be prioritized.
In this paper, we suggested a vision-based traffic accident detection algorithm and developed a system for automatically detecting, recording, and reporting traffic accidents at intersections. A system with these properties would be beneficial in determining the cause of accidents and the features of an intersection that impact safety. This model first extracts the vehicles from the video image of the charge-couple-device camera, tracks the moving vehicles (MVs), and extracts features such as the variation rate of the velocity, position, area, and direction of MVs. The model then makes decisions on the traffic accident based on the extracted features. In a field test, the suggested model achieved a correct detection rate (CDR) of 50% and a detection rate of 60%. Considering that a sound-based accident detection system showed a CDR of 1% and a DR of 66.1%, our result is a remarkable achievement
Accidents of highly automated vehicles
- A Török
- F Mammadli
Normalitásvizsgálati módszerek egy dimenzióban
- takácová
Statisztikai HipotÉzisvizsgálat
- T Móri
Normalitásvizsgálati módszerek egy dimenzióban
- N Takácová
U.S. Department of Transportation Releases Policy on Automated Vehicle Development
- Nhtsa
Machine learning for severity classification of accidents involving powered two wheelers
- N S Hadjidimitriou
- M Lippi
- M Dellamico
- A Skiera
Society of Automotive Engineers (SAE) J3016 Taxonomy and Definitions for Terms Related to Driving Automation Systems for on-Road Motor Vehicles