Road Incidents Research Articles

Detecting Unsigned Physical Road Incidents From Driver-View Images

Safety on roads is of uttermost importance, especially in the context of autonomous vehicles. A critical need is to detect and communicate disruptive incidents early and effectively. In this paper we propose a system based on an off-the-shelf deep neural network architecture that is able to detect and recognize types of unsigned (non-placarded, such as traffic signs), physical (visible in images) road incidents. We develop a taxonomy for unsigned physical incidents to provide a means of organizing and grouping related incidents. After selecting eight target types of incidents, we collect a dataset of twelve thousand images gathered from publicly-available web sources. We subsequently fine-tune a convolutional neural network to recognize the eight types of road incidents. The proposed model is able to recognize incidents with a high level of accuracy (higher than 90%). We further show that while our system generalizes well across spatial context by training a classifier on geostratified data in the United Kingdom (with an accuracy of over 90%), the translation to visually less similar environments requires spatially distributed data collection. Note: this is a pre-print version of work accepted in IEEE Transactions on Intelligent Vehicles (T-IV;in press). The paper is currently in production, and the DOI link will be added soon.

Road Data Enrichment Framework Based on Heterogeneous Data Fusion for ITS

In this work, we propose the Road Data Enrichment (RoDE), a framework that fuses data from heterogeneous data sources to enhance Intelligent Transportation System (ITS) services, such as vehicle routing and traffic event detection. We describe RoDE through two services: (i) Route service, and (ii) Event service. For the first service, we present the Twitter MAPS (T-MAPS), a low-cost spatiotemporal model to improve the description of traffic conditions through Location-Based Social Media (LBSM) data. As a case study, we explain how T-MAPS is able to enhance routing and trajectory descriptions by using tweets. Our experiments compare T-MAPS’ routes against Google Maps’ routes, showing up to 62% of route similarity, even though T-MAPS uses fewer and coarse-grained data. We then propose three applications, Route Sentiment (RS), Route Information (RI), and Area Tags (AT), to enrich T-MAPS’ suggested routes. For the second service, we present the Twitter Incident (T-Incident), a low-cost learning-based road incident detection and enrichment approach built using heterogeneous data fusion. Our approach uses a learning-based model to identify patterns on social media data which is then used to describe a class of events, aiming to detect different types of events. Our model to detect events achieved scores above 90%, thus allowing incident detection and description as a RoDE application. As a result, the enriched event description allows ITS to better understand the LBSM user’s viewpoint about traffic events (e.g., jams) and points of interest (e.g., restaurants, theaters, stadiums).

Evolution and Future of Urban Road Incident Detection Algorithms

AbstractOnly a small minority of road incident detection algorithms (IDAs) have been designed for use in urban road networks. A review of the literature conducted by the authors revealed that appro...

Statistical and machine-learning methods for clearance time prediction of road incidents: A methodology review

Accurate clearance time prediction for road incident would be helpful to evaluate the incident impacting range and provide route guiding strategy according to the predicted results, and thus reduce the travel delays caused by incidents. Currently, a number of approaches have been developed for predicting incident clearance time and investigating the effects of influential factors. Statistical and machine learning methods are the two major methodological approaches. This study aims to make a methodology review for these methods by comprehensively examining their performance in incident clearance time prediction, especially, when omitted variables present significant impacts on selected variables. Specifically, we consider four widely used statistical models: Accelerated Failure Time (AFT) model, Quantile Regression (QR) model, Finite Mixture (FM) model, and Random Parameters Hazard-Based Duration (RPHD) model, and four machine learning models: K-Nearest Neighbor (KNN) model, Support Vector Machine (SVM) model, Back Propagation Neural Network (BPNN) model, and Random Forest (RF) model as candidates. Moreover, the abilities of these methods in uncovering the underlying causality (explaining the causal effects of significant influential factors on clearance time) are also investigated. Incident clearance time data was collected on freeway road sections in Seattle, Washington State from 2009 to 2011. The conclusions can be summarized as follows: 1) the RF model and RPHD model outperform the other three models in data fitting and model prediction in their respective methodological categories; 2) three “heterogeneity” methods including RPHD, FM and QR outperform machine learning methods in model prediction as measured by MAPE; 3) machine learning methods perform stably in model prediction relative to the statistical methods; 4) incident type and lane closure type present significant effects on incident clearance time in all eight selected models.

Approaching autonomous driving with cautious optimism: analysis of road traffic injuries involving autonomous vehicles based on field test data

ObjectivesTo examine the patterns and associated factors of road traffic injuries (RTIs) involving autonomous vehicles (AVs) and to discuss the public health implications and challenges of autonomous driving.MethodsData were extracted...

Bus Arrival Time Prediction Based on LSTM and Spatial-Temporal Feature Vector

Bus arrival prediction has important implications for public travel, urban dispatch, and mitigation of traffic congestion. The factors affecting urban traffic conditions are complex and changeable. As the predicted distance increases, the difficulty of traffic prediction becomes more difficult. Forecast based on historical data responds quite slowly for changes under the short-term conditions, and vehicle prediction based on real-time speed is not sufficient to predict under long-term conditions. Therefore, an arrival prediction method based on temporal vector and another arrival prediction method based on spatial vector is proposed to solve the problems of remote dependence of bus arrival and road incidents, respectively. In this paper, combining the advantages of the two prediction models, this paper proposes a long short-term memory (LSTM) and Artificial neural networks (ANN) comprehensive prediction model based on spatial-temporal features vectors. The long-distance arrival-to-station prediction is realized from the dimension of time feature, and the short-distance arrival-to-station prediction is realized from the dimension of spatial feature, thereby realizing the bus-to-station prediction. Besides, experiments were conducted and tested based on the entity dataset, and the result shows that the proposed method has high accuracy among bus arrival prediction problems.

Recovery of Geographic Entities in Text for the Analysis of Spatial Patterns of Road Incidents

Recovery of Geographic Entities in Text for the Analysis of Spatial Patterns of Road Incidents

Problems with Not Recognising the Roadblocks at Reduced Visibility

The article was aimed at analysing the issue of legal circumstances regulating the road traffic at reduced visibility. The statistical data related to the road incidents, with special emphasis on the pedestrian accidents, were presented. The difficulties with recognising pedestrians at reduced visibility were determined.The specific roads accidents were analysed, taking into consideration their course and defining their causes. Also, the possibilities of reducing the number of the accidents with pedestrians and other roadblocks on the roadway by night were presented. The examples of real road accidents involving pedestrians included in the text include photographic materials from the accident sites and sketches prepared by police officers. The last part of the article contains a summary and formulated final conclusions.

An Effective Bio-Signal-Based Driver Behavior Monitoring System Using a Generalized Deep Learning Approach

Recent years have seen increasing utilization of deep learning methods to analyze large collections of medical data and signals effectively in the Internet of Medical Things (IoMT) environment. Application of these methods to medical signals and images can help caregivers form proper decision-making. One of the important IoMT medical application areas includes aggressive driving behaviors to mitigate road incidents and crashes. Various IoMT-enabled body sensors or camera sensors can be utilized for real-time monitoring and detection of drivers' bio-signal status such as heart rate, blood pressure, and drivers' behaviors. However, it requires a lightweight detection module and a powerful training module with real-time storing and analysis of drivers' behaviors data from these medical devices to detect driving behaviors and provides instant feedback by the administrator for safety, gas emissions, and energy/fuel consumption. Therefore, in this paper, we propose a bio-signal-based system for real-time detection of aggressive driving behaviors using a deep convolutional neural network (DCNN) model with edge and cloud technologies. More precisely, the system consists of three modules, which are the driving behaviors detection module implemented on edge devices in the vehicle, the training module implemented in the cloud platform, and the analyzing module placed in the monitoring environment connected with a telecommunication network. The DCNN model of the proposed system is evaluated using a holdout test set of 30% on two different processed bio-signal datasets. These two processed bio-signal datasets are generated from our collected bio-signal dataset by using two different time windows and two different time steps. The experimental results show that the proposed DCNN model achieves 73.02% of validation accuracy on the processed dataset 1 and 79.15% of validation accuracy on the processed dataset 2. The results confirm the appropriateness and applicability of the proposed deep learning model for detecting driving aggressive behaviors using bio-signal data.

A Framework for Enhancing the Operational Phase of Traffic Management Plans

Road traffic emergencies are dangerous and unexpected situations that require immediate actions by the authorities. These actions involve to attend to the people who have been affected by the emergency and to minimize its consequences. A Traffic Management Plan (TMP) is a set of pre-defined measures and actions designed to produce an effective and efficient use of available resources in order to deal with a specific road incident. The operational phase of a TMP involves the coordination of several independent agencies (road managers, traffic police, firemen, etc.). These agencies must provide the resources required by the TMP in the deployment of the measures and actions. In this paper, a new framework to support the TMP operational phase is presented. This framework models each agency as an intelligent agent and it uses a reverse combinatorial distributed auction as the core component of a negotiation process. The goal of this negotiation process is to obtain a common agreement on the best possible allocation of resources taking into account the role, competencies and interest of the involved agencies. The framework has been implemented in a real scenario with real data. The tests developed have demonstrated that the system is able to manage the resources in terms of the execution time and the quality of the provided solutions.

Implementation of Telematics Solutions in Urban Agglomerations in the Aspect of Road Incidents

Urban mobility is a public service provided by a road traffic management entity. The customer receives access to the road infrastructure and a service of travelling in a city by a transport means of their choice. In the case of road traffic incident management, this issue is becoming increasingly important, as every traffic management entity should deliver a product that meets road users' requirements and expectations. A characteristic element of road traffic management is incidents generated by road infrastructure users that occur at each stage of traffic management. The paper presents the results of research carried out in the aspect of use of appropriate algorithms of traffic incident management on selected national roads, supported by research and scientific discourse on aspects related to telematics systems, with particular emphasis on Intelligent Transport Systems, in order to verify the effectiveness of the implementation of telematics solutions. The issues mentioned above are extremely important in view of the need to acknowledge the expected critical infrastructure. Principles and recommendations used in the selection and implementation of ITS applications become an important element in this respect. The research was used to verify the effectiveness of event management algorithms in road traffic, with different traffic volume and meteorological conditions. Empirical findings used in research allow for the analysis of changes in traffic parameters, such as vehicle speed, traffic volume and detector occupancy, on selected national roads, at specific intervals. This has made it possible to determine the prospects for the development of traffic incident management algorithms, which constitute a set of artificial intelligence methods.

MuTraff: A Smart-City Multi-Map Traffic Routing Framework.

Urban traffic routing is deemed to be a significant challenge in intelligent transportation systems. Existing implementations suffer from several intrinsic issues such as scalability in centralized systems, unnecessary complexity of mechanisms and communication in distributed systems, and lack of privacy. These imply force intensive computational tasks in the traffic control center, continuous communication in real-time with involved stakeholders which require drivers to reveal their location, origin, and destination of their trips. In this paper we present an innovative urban traffic routing framework and reference architecture (multimap traffic control architecture, MuTraff), which is based on the strategical generation and distribution of a set of traffic network maps (traffic weighted multimaps, TWM) to vehicle categories or fleets. Each map in a TWM map set has the same topology but a different distribution of link weights, which are computed by considering policies and constraints that may apply to different vehicle groups. MuTraff delivers a traffic management system (TMS), where a traffic control center generates and distributes maps, while routing computation is performed at the vehicles. We show how this balance between generation, distribution, and routing computation improves scalability, eases communication complexities, and solves former privacy issues. Our study presents case studies in a real city environment for (a) global congestion management using random maps; (b) congestion control on road incidents; and c) emergency fleets routing. We show that MuTraff is a promising foundation framework that is easy to deploy, and is compatible with other existing TMS frameworks.

Seasonal patterns of traumatic brain injury deaths due to traffic-related incidents in the Slovak Republic

Objective: Traumatic Brain Injuries (TBIs) are an important type of injury in terms of both morbidity and mortality. Road Traffic Incidents are one of the most frequent causes of TBI. This analysis seeks to quantify the number of such injuries occurring in the Slovak Republic, and examine patterns of TBI according to mode of transport and seasonality.Methods: Data concerning total numbers of TBIs occurring from the years 1996–2015 were obtained from the Statistical Office of the Slovak Republic. The events caused by road incidents were examined separately according the external cause stated on death certification. Events were classified into seasons according to the month of death. Summary statistics were produced concerning numbers of deaths according to sex, mode of transport and season. Analyses were performed to examine trends in TBI by season and type of road user.Results: During a period of 20 years from 1996, there were 17,047 recorded deaths involving TBI in the Slovak Republic. Of these, 5,370 were caused by road traffic incidents (RTIs). Age standardized rates tended to decrease from 8.3/100,000/year (1996) to 2.5/100,000/year (2015). Males made up approximately 79% of road traffic-caused TBIs. Summer and autumn showed significantly more events than any other season, with motorcyclists and cyclists in particular being more frequently injured at this time of year.Conclusions: The results show that Slovakia, like many countries, suffers a considerable burden of TBI and that RTIs are a major contributor to this, especially among young adults. Rates of TBI vary by season in Slovakia, and users of different modes of transport appear more or less likely to suffer such injury during different seasons. Considerable variability in rates of injury exists between road users and times of year. Improved understanding of the timing and sufferers of injuries may allow better planning of response and care services. Further research into transport modes and policies aimed at safer driving should be explored.

A review on impact of human factors on road safety with special focus on hazard perception and risk-taking among young drivers

Road traffic accidents represented as the third cause of “death and disease” also the major cause of death among young people aged 15-29 years. In recent decade consequential investigation has been conducted on road crash contributing factors. All these researches showed that the recurrence and severity of accidents depend on many causes, such as vehicle performance, roadway environment, and characteristics of road users. Despite many studies in this field, the function of human factors on road accidents, is still unclear, due to the complex human characteristics. The lack of comprehensive model or procedure for scrutiny and evaluating the human characteristics which contributed to road incidents, motivate us to review on previous studies in order to indicate the relationships between human factors and road safety. This paper focus on the natural driving behavior of drivers including hazard perception and risk-taking of drivers which is affected by age, perceptual/cognitive abilities and driving experience. The study clearly shows that the mentioned factors have a huge impact on road incidents. Therefore, this paper indicate and recommend a number of solutions in order to mitigate the severity and frequency of accidents caused by human errors. Also previous road safety studies will provide series of crucial information for improvement of driver licensing and their training procedure.

Vehicle and Highway Adaptations to Compensate for Sleepy Drivers.

Sleepiness remains a major contributor to road crashes. Driver monitoring systems identify early signs of sleepiness and alert drivers, using real-time analysis of eyelid movements, EEG activity, and steering control. Other vehicle adaptations warn drivers of lane departures or collision hazards, with higher vehicle automation actively taking over vehicle control to prevent run off the road incidents and institute emergency braking. Similarly, road adaptations warn drivers (rumble strips) or mitigate crash severity (barriers). Infrastructure to encourage drivers to use countermeasures, such as rest stops for napping, is also important. The effectiveness of adaptations varies for different road users.

Road safety in Poland Three Years After Limiting the Number of Speed Controls

Road safety is one of the key elements of every citizen’s life. Every day journeys connected with professional and educational activity force us to do a trip, and this leads to inevitable conflicts between road users. One of the basic methods of assessing the state of road safety, and the effects of legal, educational and infrastructural activities of authorities, are the results of road incidents analyses, accidents and collisions, as well as the effects of these activities in previous years. Police databases carried out in most countries on a continuous basis, store records of road incidents in which Police officers intervened. Undoubtedly the speed is one of the most important element affecting the number and consequences of road incidents. High speed reduces the time necessary for the driver to react correctly to the traffic situation. The high speed increases the momentum, and thus the effects of road accidents. Changes in the scope of speed control of vehicles in Poland, including those entitled to do it, introduced 3 years ago, could significantly affect the state of road safety in Poland, on the network of urban and non-urban roads. This paper presents the results of road safety analyses for both nationally and more locally road networks - for the biggest cities in Poland. Based on the analysis of data from the Police’s database records, it has been shown that since the introduction of legal changes, the number of road incidents on the road network has increased significantly. It also indicates unfavourable trends in the area of road safety in cities. Based on the above analyses the need for action to introduce a new program for road safety improvement, especially at the level of urban areas, is indicated.

Data‐driven next destination prediction and ETA improvement for urban delivery fleets

In an urban logistics system, predictions of the next destination and estimated time of arrival (ETA) are of paramount importance for efficient resource planning of delivery fleets and for providing a satisfactory client experience. The quality of prediction is limited by the information accessible to individual logistics business entities, and further complicated by the complex urban road system. Data collection under the auspices of smart city initiatives worldwide provides exciting new opportunities to overcome these limitations. In this study, the authors identify two areas of improvement through data-driven approaches, including a next destination predictor, based on the delivery fleet's historical global positioning system trajectory data using a non-linear autoregressive neural network, and a road incident detector for real-time ETA improvement. By comparing a range of machine learning classification algorithms for incident detection, XGBoost has been found to be the most practical choice, due to its performance and efficiency. The proposed framework can be utilised by government authorities who possess such data for better urban planning and providing advanced infrastructure, so as to improve the operational efficiency of the logistics industry.

Optimisation of lane-changing advisory at the motorway lane drop bottleneck

Reduction in the number of lanes (lane drop) is common on motorways due to road design, incidents, or road maintenance, and it can be an active bottleneck if the traffic demand is high. If congestion occurs, the lane drop capacity will decrease 10–20%. In order to avoid capacity drop, this study analysed the reason and proposed lane-changing advisory control on the merge lane to distribute lane-changing using Cooperative Intelligent Transport Systems (C-ITS) technology. Further, this study used hyper-heuristic optimisation to obtain the lane-changing advisory proportion of each segment upstream of lane drop. Conditions of different traffic demands were analysed using the microscopic traffic simulation software AIMSUN and its Application Program Interface (API) function. Results indicated that the proposed lane-changing advisory strategy could reduce traffic congestion and obviously improve traffic efficiency. This study also analysed the effects with different proportions of connected vehicles and found that if the connected vehicle ratio is less than 20%, the lane-changing advisory has little impact on the lane drop performance, and that if the penetration rate is more than 50%, the gain in performance is marginal.

Young drivers who continue to drive while sleepy: What are the associated sleep- and driving-related factors?

Crashes due to sleepiness account for a substantial proportion of road crash incidents. The purpose of the current study was to examine several sleep-related factors and driving-related factors for their association with self-reports of continuing to drive while sleepy. In total 257 young drivers aged 18-25years completed an online survey that assessed factors such as sleep quality, sleep duration and consistency, excessive daytime sleepiness, experiences with sleepiness and their driving-related behaviours. The results demonstrate that being older, having a perceived ability to overcome sleepiness, committing more highway code violations and having experienced a sleep-related close call were positively associated with an increased likelihood of continuing to drive while sleepy. The obtained results highlight the acceptance of risky driving behaviours among some younger drivers. Younger drivers' risky driving behaviour is certainly a road safety concern given the impairment associated with sleepiness and their over-representation in road crash incidents.

Intelligent monitoring system for smart road environment

Abstract The article considers the results of research received by the authors in the process of developing and implementing the components of an intelligent monitoring system. The system is designed to collect and process big data on road incidents from photo-radar complexes. Monitoring system is the main part of the Smart Road Environment. The authors propose a convergent model for processing big sensor data, which includes three layers using fog, cloud and mobile computing technologies. Within the framework of the convergent model, the multi-agent approach for data processing is implemented. In the process of working on the project, software agents that perform data collecting, cleaning, clustering, comparing time series, retrieving data for visualization, preparing charts and reports, performing spatial analysis, etc. Big sensor data and other heterogeneous information from other sources (weather stations, video surveillance systems, mobile communication equipment, built-in car navigation equipment etc.) are collected. The results of data processing mining are presented. The comparison of time series of traffic accidents and time series of meteorological factors is established, that the presence of factors to become determinants for an abnormal change in the traffic situation in controlled areas.