Highway Environment Research Articles

Deep learning-based UAV image segmentation and inpainting for generating vehicle-free orthomosaic

• A novel deep learning-based image segmentation and inpainting method was proposed to generate vehicle-free road orthomosaic. • This method is removing the unwanted objects from each original images captured by UAV before processing the orthomosaic. • By comparing the proposed and traditional ways, this study validated the performance of the UAV image inpainting to generate the vehicle-free orthomosaic. • The proposed method can remove provide contextual information more effectively by eliminating the unwanted objects. A low-altitude orthomosaic derived by an unmanned aerial vehicle (UAV) has been widely utilized for various purposes in large-scale infrastructure management. However, unwanted objects, such as cars and trucks, captured in the aerial images captured by the UAV have negative impacts on the quality of orthomosaic. To this end, this study presented a novel method to remove the effect of unwanted objects on UAV-generated orthomosaic. The proposed method applied a deep learning-based image segmentation and inpainting algorithm to remove the vehicles from individual UAV images before processing structure from motion (SfM), and then it resulted in generateing an orthomosaic with the inpainted UAV images. To validate the proposed method, this study conducted a case study in actual highway environment and compared the performance of the proposed method with that of another method, which directly removes and inpaints vehicles from the final orthomosaic. Through comparison tests, it is shown that the proposed method is more effective than the other. The proposed automatic vehicle-free orthomosaic generation method can contribute to creating up-to-date immersive content for transportation infrastructure management.

Parallelized and randomized adversarial imitation learning for safety-critical self-driving vehicles

Self-driving cars and autonomous driving research has been receiving considerable attention as major promising prospects in modern artificial intelligence applications. According to the evolution of advanced driver assistance system (ADAS), the design of self-driving vehicle and autonomous driving systems becomes complicated and safety-critical. In general, the intelligent system simultaneously and efficiently activates ADAS functions. Therefore, it is essential to consider reliable ADAS function coordination to control the driving system, safely. In order to deal with this issue, this paper proposes a randomized adversarial imitation learning (RAIL) algorithm. The RAIL is a novel derivative-free imitation learning method for autonomous driving with various ADAS functions coordination; and thus it imitates the operation of decision maker that controls autonomous driving with various ADAS functions. The proposed method is able to train the decision maker that deals with the LIDAR data and controls the autonomous driving in multi-lane complex highway environments. The simulation-based evaluation verifies that the proposed method achieves desired performance.

ESG-oriented Governance Work System for Smart Highway Implementation

This study was promoted to prepare an ESG-oriented work system for Korea Expressway Corporation through sustainable highway construction, IOT-based smart highway preparation, and establishment of a broad convergence business governance system between multiple ministries and implementing entities. Based on the 2nd Basic Plan for Landscape Policy (2020~2024) and the Public Design Promotion Act (2018~2022), the study was conducted through the improvement of the highway design method to be applied for four years from 2021 to 2024, the changes in construction and social conditions related to expressways, the analysis of policies and plans to secure consistency with the 'Korean New Deal', and the analysis of the project performance and work promotion system that the Korea Expressway Corporation's business execution organization has promoted together with each government department. The results of this research are expected to help the Korea Expressway Corporation establish the 21st Innovation Plan for the phased construction of a “100-years landscape with people” and the preparation of a sustainable future highway environment creation project for the lives and safety of the people.

Road infrastructures, spatial surroundings, and the demand and route choices for cycling: Evidence from a GPS-based mode detection study from Oslo, Norway

To achieve a higher cycling uptake, it is essential for planners to know what kind of cycling infrastructure to plan and where, that is, through which types of urban environments. In this paper, we provide a deeper understanding of cycling demand and cycling route choices and infer insights into cyclists’ latent preferences and dispreferences concerning both infrastructure attributes and the spatial characteristics of route surroundings. Hereto, this study has collected, map-matched, geovisualized, and examined a unique GPS-based database with over 25,915 cycling trips in Oslo, Norway. Our findings reveal that cyclists substantially deviate from shortest paths, covering 59% more distance on average. Higher cycling frequencies, both in absolute terms and relative to shortest-path-expected-values, can be found on route sections that have some form of cycling infrastructure, especially those having segregated bicycle highways and bike roads. We also find higher demand and route choices for flatter and water-facing routes, as well as routes less disrupted by crossings and away from highway environments. In contrast, routes surrounded by green space or high population density, despite having high demand in absolute terms, are cycled less than expected based on shortest paths. The paper concludes by reflecting on the significance, limitations, and implications of our findings and novel methodological approaches for the bicycle route choice theory and practice moving forward.

Concentrations of tire wear microplastics and other traffic-derived non-exhaust particles in the road environment

Tire wear particles (TWP) are assumed to be one of the major sources of microplastic pollution to the environment. However, many of the previously published studies are based on theoretical estimations rather than field measurements. To increase the knowledge regarding actual environmental concentrations, samples were collected and analyzed from different matrices in a rural highway environment to characterize and quantify TWP and other traffic-derived non-exhaust particles. The sampled matrices included road dust (from kerb and in-between wheeltracks), runoff (water and sediment), and air. In addition, airborne deposition was determined in a transect with increasing distance from the road. Two sieved size fractions (2–20 µm and 20–125 µm) were analyzed by automated Scanning Electron Microscopy/Energy Dispersive X-ray spectroscopy (SEM/EDX) single particle analysis and classified with a machine learning algorithm into the following subclasses: TWP, bitumen wear particles (BiWP), road markings, reflecting glass beads, metals, minerals, and biogenic/organic particles. The relative particle number concentrations (%) showed that the runoff contained the highest proportion of TWP (up to 38 %). The share of TWP in kerb samples tended to be higher than BiWP. However, a seasonal increase of BiWP was observed in coarse (20–125 µm) kerb samples during winter, most likely reflecting studded tire use. The concentration of the particle subclasses within airborne PM80-1 decreases with increasing distance from the road, evidencing road traffic as the main emission source. The results confirm that road dust and the surrounding environment contain traffic-derived microplastics in both size fractions. The finer fraction (2–20 µm) dominated (by mass, volume, and number) in all sample matrices. These particles have a high potential to be transported in water and air far away from the source and can contribute to the inhalable particle fraction (PM10) in air. This highlights the importance of including also finer particle fractions in future investigations.

V2V channel modeling at 5.2 GHz for highway environment

To design and evaluate vehicle-to-vehicle (V2V) communication systems in intelligent transportation system (ITS), it is important to understand the propagation mechanisms and channel models of V2V channels. This paper aims to analyze the channel models at 5.2 GHz for the highway environment in obstructed line-of-sight (OLoS) and line-of-sight (LoS) scenarios, particularly the vehicle connectivity probability derivation based on the propagation model obtained from measurement. First, the path loss (PL), shadow fading (SF), narrowband K-factor, and small-scale amplitude fading are analyzed. Results showed that the received signal magnitude follows Rice and Weibull distribution in LoS and OLoS scenarios, respectively. Second, we develop simple and low-complexity tapped delay line (TDL) models with a 10 MHz bandwidth for LoS and OLoS scenarios; in addition, we investigate the wideband K-factor, the root mean square delay spread (RMS-DS), and delay-Doppler spectrum. Third, we derive the closed form connectivity probability between any two vehicles in the presence of Weibull fading channel, and analyze the effects of Weibull fading channel and traffic parameters on connectivity. It is found that Weibull fading parameter, transmit power and vehicle density have positive impact on connectivity probability, PL exponent has negative impact on connectivity probability.

Post-assessment of the eco-environmental impact of highway construction– A case study of Changbai Mountain Ring Road

The eco-environmental impact of highway construction is a long-term and cumulative process. Most of the existing studies pay attention to the environmental impact assessment technology of highway construction period and completion acceptance period, which can not fully explain the eco-environmental impact of highway operation period. At the same time, most of the previous studies have a single index and less consideration of the impact on species. Aiming at the highway located in the forest ecosystem in the middle temperate zone, considering the four aspects of species, land use, landscape pattern and ecological function, this paper establishes the post-evaluation index system of the eco-environmental impact of highway construction projects. This paper puts forward the quantitative assessment technology and spatio-temporal scale identification method of the long-term evolution of the impact of highway construction on the ecological environment, and uses the analytic hierarchy process to construct a comprehensive post-assessment system of ecological environmental impact of highway construction projects. Relying on the Changbai Mountain Ring Road, a 10-year comprehensive ecological impact assessment was carried out. Mammals tend to accumulate within 200 m of the road, and highway corridors attract bird activities; the animal mortality coverage rates of amphibians and reptiles are 100%, mammals 66.7%, and birds 50.9%. The utilization rate of bridges and culverts is 88% and 42%, and the total utilization rate is 49.3%. The landscape is changing to the dominant type of coniferous and broad-leaved mixed forest, with lower evenness, less diversity and higher degree of natural connection, but the ability of the ecological environment to resist external changes is reduced. The closer the landscape fragmentation is to the highway, the more serious it is, and the lower the diversity is. The time and space scale of post-assessment of eco-environmental impact of highway construction project is in the range of 4–7 years and 200-300 m on the road side, and with the increase of operation time, the change of ecosystem along the line decreases year by year and tends to be stable gradually. The research results can provide a method of reference for the follow-up study of highway ecological environment in mid-temperate forest ecological areas.

Congestion aware low power on chip protocols with network on chip with cloud security

This article is to analyze the bottleneck problems of NoC in many more applications like multi-processor communication, computer architectures, and network interface processors. This paper aims to research the advantages and disadvantages of low congestion protocols on highway environments like multiple master multiple slave interconnections. A long-term evolution and effective on-chip connectivity solution for secured, congestion aware and low power architecture is emerged for Network-on-Chip (NoC) for MCSoC. Applications running simultaneously on a different chip are often exchanged dynamically on the chip network. Of-course, in general on chip communication, resources mean that applications may interact with shared resources to influence each other's time characteristics.

A novel extended Kalman filter-based optimized routing approach for IoV environment

The development of intelligent transportation system has divulged towards traffic management, planning and control that requires precise neighboring vehicle location anticipation for information transmission. The crucial problem is to maintain QoS parameters in high speed and varying vehicular topology environment. The paper presents a novel EK-PGRP (Extended Kalman filter- Predictive Geographic Routing Protocol) routing approach to anticipate neighbor location and to select the propitious neighbor for advancing packets from source to destination vehicle using extended Kalman filter for real-time V2V communication in both urban and highway vehicular environment. This is acquired according to spatial and temporal movement attributes; every vehicle has an anticipation model to anticipate its own and neighboring vehicle mobility. Moreover, if LMP (Local Maximum Problem) state is encountered, i.e., where a vehicle is unable to locate any neighbor nearer to destination than itself to forward an information packet; then it uses predictive prediction algorithm to overcome that state. The precision, robustness and coherence attributes of the proposed routing approach are illustrated via extensive simulations. EK-PGRP is contrasted with K-PGRP (Kalman filter- Predictive Geographic Routing Protocol), PGRP (Predictive Geographic Routing Protocol) and GPSR (Greedy Perimeter Stateless Routing) routing protocols and results demonstrate that EK-PGRP outperformed most of the simulation cases and attained the minimum location error while ameliorating prediction accuracy of the vehicles in vehicular environment. The simulations were performed on MATLAB R2018a along with traffic simulator SUMO.

Partially automated driving has higher workload than manual driving: An on‐road comparison of three contemporary vehicles with SAE Level 2 features

AbstractVehicles with SAE Level 2 automated features are already in active use on the road, and vehicles with Level 3 or 4 will be with us soon. Although the vehicles provide support for longitudinal and lateral control, partially automated driving experience is sometimes more demanding than manual driving. However, the effects of automated driving on workload in naturalistic conditions have not been extensively investigated, as most studies have been undertaken in driving simulators. This study aims to extend the current understanding about workload in partially automated driving on public roads. Drivers' perceived workload was assessed after conducting manual and automated driving activities using a small sample (N = 8). They performed driving tasks in three contemporary vehicles with SAE Level 2 features, in highway and urban environments. The comparative findings revealed that drivers' perceived workload was higher in partially automated driving than manual driving. Furthermore, perceived workload was higher in urban environments than highway environments and in less experienced drivers than more experienced drivers. Although the findings may need to be interpreted with caution due to the small sample size, they provide a future research agenda that can be built upon.

Research on driving visual comfort of two‐lane mountainous road based on traffic safety evaluation

AbstractIn order to obtain the quantitative relationship between multisource data from road landscape and driving visual comfort in two‐lane mountainous road, we collected the driver's eye movement data and road environment data through real vehicle tests. The threshold segmentation method and the target landscape area ratio method are used to process the visual image, and the spatial enclosure degree is used to express the driver's sense of spatial closure. In addition, an HSV color model is established to characterize the driver's visual perception of the mountain highway landscape environment. Combining the target landscape area ratio of road landscape with the amount of color information, the calculation method of road landscape information is obtained. According to the characteristics of normal distribution of pupil area ratio, five grades of driver's visual comfort level are proposed, and the regression relationship between comfort level and landscape information is established. The study found that there is a significant negative correlation between the amount of road landscape information and the driver's visual comfort. When the amount of road landscape information is less than 0.338, the driving comfort is higher, which is conducive to driving safety. However, when the amount of landscape information in the road area exceeds 0.644, the driving comfort decreases significantly, which is detrimental to driving safety.

Policy-Based Reinforcement Learning for Training Autonomous Driving Agents in Urban Areas With Affordance Learning

Learning to drive in urban areas is an open challenge for autonomous vehicles (AVs), as complex decision making requirements are needed in multi-task co-ordinations environments. In this paper, we propose a hybrid framework with a new perception model involving affordance learning to simplify the surrounding urban scenes for training an AV agent, along with a planned trajectory and the associated driving measurements. Our proposed solution encompasses two main aspects. Firstly, a supervised learning network is used to map the input sensory data into affordance predictions. The predicted affordances provide a low-dimensional representation of surrounding scenes of the AV in the form of key perception indicators, e.g., true or false with respect to a traffic light signal. Secondly, a deep deterministic policy gradient model that maps the perception information into a series of actions is devised. We evaluate the proposed solution using the CARLA driving simulator in an urban town and evaluate the performance in a new, unseen town under different weather conditions. The quantitative and qualitative results indicate that our proposed solution can generalize well to cope with different traffic solutions and environmental conditions. Our proposed solution also outperforms other baseline methods in a comparative study in handling various AV driving tasks with different levels of difficulty. In addition, the model trained with simulated scenes yields promising prediction results when testing on recorded video streams on real-world highway and suburb environments with varying traffic and weather conditions.

A Unified Framework Integrating Decision Making and Trajectory Planning Based on Spatio-Temporal Voxels for Highway Autonomous Driving

Intelligent decision making and efficient trajectory planning are closely related in autonomous driving technology, especially in highway environment full of dynamic interactive traffic participants. This work integrates them into a unified hierarchical framework with long-term behavior planning (LTBP) and short-term dynamic planning (STDP) running in two parallel threads with different horizon, consequently forming a closed-loop maneuver and trajectory planning system that can react to the dynamic environment effectively and efficiently. In LTBP, a novel voxel structure and the ‘voxel expansion’ algorithm are proposed for the generation of driving corridors in 3D configuration, which involves the prediction states of surrounding vehicles. By using Dijkstra search, the maneuver with minimal cost is determined in form of voxel sequences, then a quadratic programming (QP) problem is constructed for solving the optimal trajectory. And in STDP, another small-scaled QP problem is performed to track or adjust the reference trajectory from LTBP in response to the dynamic obstacles. Meanwhile, a Responsibility-Sensitive Safety (RSS) Checker keeps running at high frequency for real-time feedback to ensure security. Experiments on real data collected in different highway scenarios demonstrate the effectiveness and efficiency of our work.

Machine learning based real-time prediction of freeway crash risk using crowdsourced probe vehicle data

Real-time prediction of crash risk can support traffic incident management by generating critical information for practitioners to allocate resources for responding to anticipated traffic crashes proactively. Unlike previous studies using archived traffic data covering a limited highway environment such as a segment or corridor, this study uses a statewide live traffic database from HERE to develop real-time traffic crash prediction models. This database provides crowdsourced probe vehicle data that are high-resolution real-time traffic speed for the entire freeway network (nearly 2,000 miles) in Alabama. This study aims to use machine learning models to predict crash risk on freeways according to pre-crash traffic dynamics (e.g., mean speed, speed reduction) along with static freeway attributes. Traffic speed characteristics were extracted from the HERE database for both pre-crash and crash-free traffic conditions. Random Forest (RF), Support Vector Machine (SVM) and Extreme Gradient Boosting (XGBoost) were developed and compared. Separate models were estimated for three major crash types: single-vehicle, rear-end, and sideswipe crashes. The model prediction accuracy indicated that the RF models outperform other models. Models for rear-end crashes are found to have greater accuracy than other models, which implies that rear-end crashes have a significant relationship with pre-crash traffic dynamics and are more predictable. The traffic speed factors that are ranked high in terms of feature importance are the speed variance and speed reduction prior to crashes. According to partial dependence plots, the rear-end crash risk is positively related to the speed variance and speed reductions. More results are discussed in the paper.

Experimental V2X Evaluation for C-V2X and ITS-G5 Technologies in a Real-Life Highway Environment

In the coming years, connectivity between vehicles with autonomous driving features and roadside infrastructure will become more and more a reality on our roads, pursuing to improve road safety and traffic efficiency. In this regard, two main communication standards are considered as key enablers, that is ITS-G5 (based on IEEE 802.11p) and C-V2X (3GPP). To assess the real performance of these technologies, there is still need for an objective and independent one-to-one comparison of these technologies using off-the-shelf hardware under identical and real-life traffic conditions. Until today, performance evaluations are limited to simulations, emulations or individual technology assessments in real-life circumstances. In this paper, an exhaustive and fair evaluation of the technologies has been conducted in a real-life highway environment under identical conditions. Tailored evaluation tools in combination with our in-house CAMINO vehicular framework has been utilized to perform the tests and analyze the results for different well-specified test cases. The performance evaluation shows that for the short-range technologies, C-V2X PC5 has, in general, a higher range than ITS-G5, while ITS-G5 offers lower latency than C-V2X PC5 in low-density scenarios. Long-range 4G C-V2X can be considered as an alternative for certain use cases. The outcome of this experimentation study can be used as valuable information for the further development of future (5G) connected and autonomous driving.

Establishment of Performance Evaluation Model of Highway Environmental Protection Based on Deep Learning

As the foundation of the social economic system, public transportation roads play an important role in the development of the national economy. At the same time, the opposition between public transportation roads and the natural environment, ecological environment, and natural resources is becoming more and more obvious. This paper is aimed at studying the performance evaluation model of highway environmental protection based on deep learning. For this reason, related methods such as environmental protection, performance evaluation, and deep learning algorithms are proposed in the article, and corresponding experiments are also carried out in the article. The experimental results show that the highway environmental protection performance evaluation model based on deep learning can quickly help environmental supervisors to inspect the work of relevant highway environmental protection agencies and also accurately mention the environmental protection performance detection to 80%.

Security Improvements for Connected Vehicles Position-Based Routing

The constant growing on the number of vehicles is increasing the complexity of traffic in urban and highway environments. It is paramount to improve traffic management to guarantee better road usage and people’s safety. Through efficient communications, Vehicular Ad hoc Networks (VANETs) can provide enough information for traffic safety initiatives, daily traffic data processing, and entertainment information. However, VANETs are vulnerable to malicious nodes applying different types of net-work attacks, where an attacker can, for instance, forge its position to receive the data packet and drop the message. This can lead vehicles and authorities to make incorrect assumptions and decisions, which can result in dangerous situations. Therefore, any data dissemination protocol designed for VANET should consider security issues when selecting the next-hop forwarding node. In this paper, we propose a security scheme designed for position-based routing algorithms, which analyzes nodes position, transmission range, and hello packet interval. The scheme deals with malicious nodes performing network attacks, faking their positions forcing packets to be dropped. We used the Simulation of Urban MObility (SUMO) and Network Simulator-version 3 (NS-3) to compare our proposed scheme integrated with two well-known position-based algorithms. The results were collected in an urban Manhattan grid environment varying the number of nodes, the number of malicious nodes, as well as the number of source-destination pairs. The results show that the proposed security scheme can successfully improve the packet delivery ratio while maintaining low average end-to-end delay of the algorithms.

Fully Automated Algorithm for Light Pole Detection and Mapping in Rural Highway Environment Using Mobile Light Detection and Ranging Point Clouds

The widespread use of light detection and ranging (LiDAR) data provides a promising source for the automatic detection and inventory of roadside assets. One of the essential elements in roadside furniture is light poles. There is limited research on the mapping of light poles using point-cloud data on rural highways. In this environment, the placement of light poles within roadside clear zones often poses a safety concern, as they are related to an increased risk of collisions. Only a limited number of studies have explored the relationship between light poles and safety because of the time-consuming and labor-intensive practices of collecting light pole assets data using traditional manual methods. This paper proposes an automated approach to mapping the locations of light poles. First, the scanning vehicle trajectory is extracted, smoothed, and then used to segment the point-cloud data into smaller overlapped batches of data. Several filters are applied to extract pole-like objects from the data. The segments are combined back together, and a density-based clustering algorithm is used to group the remaining points into clusters. A geometric filter is finally applied to extract light poles. The model is tested on 28 km of data on three rural highways in Alberta, Canada. The proposed algorithm is found to be accurate relative to previous studies, with average precision, recall, and F1 scores exceeding 98% for the test segments. The proposed work can assist in the automation of light pole inventory and road safety audits by transportation agencies.

Multi-agent reinforcement learning for cooperative lane changing of connected and autonomous vehicles in mixed traffic

Autonomous driving has attracted significant research interests in the past two decades as it offers many potential benefits, including releasing drivers from exhausting driving and mitigating traffic congestion, among others. Despite promising progress, lane-changing remains a great challenge for autonomous vehicles (AV), especially in mixed and dynamic traffic scenarios. Recently, reinforcement learning (RL) has been widely explored for lane-changing decision makings in AVs with encouraging results demonstrated. However, the majority of those studies are focused on a single-vehicle setting, and lane-changing in the context of multiple AVs coexisting with human-driven vehicles (HDVs) have received scarce attention. In this paper, we formulate the lane-changing decision-making of multiple AVs in a mixed-traffic highway environment as a multi-agent reinforcement learning (MARL) problem, where each AV makes lane-changing decisions based on the motions of both neighboring AVs and HDVs. Specifically, a multi-agent advantage actor-critic (MA2C) method is proposed with a novel local reward design and a parameter sharing scheme. In particular, a multi-objective reward function is designed to incorporate fuel efficiency, driving comfort, and the safety of autonomous driving. A comprehensive experimental study is made that our proposed MARL framework consistently outperforms several state-of-the-art benchmarks in terms of efficiency, safety, and driver comfort.

Development and application of an innovative approach to predicting pollutant concentrations in highway runoff

Recognising the challenges and limitations of current methodologies to predict highway runoff concentrations, this paper presents a novel approach based on the derivation of pollutant emission factors for twelve different types of vehicle. Published emission factor data and properties of differing vehicles types are combined with annual average daily traffic volume (AADT), highway characteristics and rainfall data to determine the pollutant distributions associated with differing highway and traffic types. In this paper, the method is applied to 126 sections of highway in the Greater London Borough of Enfield (United Kingdom; UK) and results are comparable with values reported in the literature. The approach is used to identify the level of AADT predicted to result in an exceedance of environmental quality standards (EQS), with results suggesting that runoff from highways experiencing AADT values as low as 5000 may require treatment prior to discharge to receiving waters. Future scenario analyses indicate that the impact of progressively replacing petrol and diesel vehicles with electric vehicles will have negligible impact on concentrations of zinc (Zn), copper (Cu), cadmium (Cd) and total suspended solids discharging from highway environments. The approach enables identification and ranking of urban highways in terms of their pollution runoff potential and provides an important support to users in prioritising locations for the installation of sustainable drainage options in order to protect receiving water environments.