Are Work Zones and Connected Automated Vehicles Ready for a Harmonious Coexistence? A Scoping Review and Research Agenda
DDehman and Farooq 1
Are Work Zones and Connected Automated Vehicles Ready for a Harmonious Coexistence? A Scoping Review and Research Agenda
Amjad Dehman, Ph.D., Senior Researcher*
Laboratory of Innovations in Transportation (Litrans) Centre for Urban Innovation, Ryerson University 44 Gerrard Street East, Toronto ON M5G 1G3, Canada Emails: [email protected], [email protected] Phone: 416-979-5000 ext. 556456
Bilal Farooq, Ph.D., Associate Professor
Laboratory of Innovations in Transportation (Litrans) Centre for Urban Innovation, Ryerson University 44 Gerrard Street East, Toronto ON M5G 1G3, Canada Emails: [email protected] Phone: 416-979-5000 ext. 556456 * Corresponding Author
Abstract
The recent advent of connected and automated vehicles (CAVs) is expected to transform the transportation system. CAV technologies are being developed rapidly and they are foreseen to penetrate the market at a rapid pace. On the other hand, work zones (WZs) have become common areas on highway systems as a result of the increasing construction and maintenance activities. The near future will therefore bring the coexistence of CAVs and WZs which makes their interaction inevitable. WZs expose all vehicles to a sudden and complex geometric change in the roadway environment, something that may challenge many of CAV navigation capabilities. WZs however also impose a space contraction resulting in adverse traffic operational and safety impacts, something that legitimately calls for benefiting from the highly efficient CAV mobility and safety functions. CAVs should be able to reliably traverse WZ geometry and WZs should benefit from CAV intelligent functions. This paper explores the key concepts of deploying CAV systems at WZs with a focus on mobility, safety, and infrastructure considerations. CAV concepts were distributed over the WZ area which was subdivided into five segments: further upstream, approach area, queuing area, WZ activity, and termination area. The paper also provides a review on the recent literature and summarizes a research agenda. The paper aims to provide a bird’s eye view, but with necessary details that can benefit practitioners, planners, and transportation agencies. Existing findings show that CAVs can credibly solve many WZ mobility and safety challenges, CAV benefits however are very sensitive to market penetration rate, traffic volume, communication technologies, and the assumed driver behaviour. A plenty of research work has yet to be organized and conducted in order to better harmonize the forthcoming coexistence between CAVs and WZs.
Keywords:
Connected and automated vehicles, work zone, traffic operations, traffic flow efficiency, traffic safety, traffic control devices. ehman and Farooq 2 INTRODUCTION
Connected and Automated Vehicles (CAVs) are being developed rapidly and they are foreseen to considerably penetrate the market in the near future. Their advanced and smart functions can remarkably improve highway mobility and safety. On the other hand, work zones (WZs) have become common areas on the highway system in response to the growing construction and maintenance activities. The coexistence of CAVs and WZs have so become inevitable. The interesting question would be “Is this coexistence an opportunity, a challenge, or both?”
WZs may benefit from CAV smart functions to improve traffic flow performance, CAVs however need to accurately perceive the complex geometry of WZs and traverse these areas reliably. Statistics have been increasingly and continuously stressing on the impact of work zones on mobility and safety. The US Federal Highway Administration (FHWA) [1] estimated that construction causes around 10% of the total road delay in the U.S., construction comes after recurring bottlenecks (40%), traffic incidents (25%), and inclement weather (15%). This 10% work-zone-induced congestion share can cause an enormous financial impact knowing that the overall traffic congestion cost is substantially high. A study by Metrolinx and HDR Corporation [2] estimated that the cost of congestion in the Greater Toronto and Hamilton Area is around $6 billion annually based on the 2006 travel data, and this figure is forecasted to increase to $15 billion per year by 2031. The annual cost of traffic congestion in the US was estimated to be around $179 billion based on 2017 travel data covering 494 U.S. urban areas [3]. Beside mobility impacts, WZs also create significant safety challenges. According to FHWA statistics, work zones resulted in 587 traffic-related fatalities in 2011 and 37,000 injuries in 2010; also, an average of two people are killed and 101 are injured every day in U.S. highway work zones [4]. The adverse WZ-induced mobility and safety impacts legitimately call for investing heavily in advanced technologies and smart systems. With the looming CAV penetration in the highway network, benefiting from the highly efficient CAV mobility and safety functions at WZs is expected to be an active and important field of research. Using the Transport Research International Documentation (TRID), an integrated database that combines more than 1.25 million records of transportation-related research worldwide, Figure 1 demonstrates the evolution of scientific publications relevant to both CAVs and WZs in the last three decades. The used search keywords were “work zone” and “connected and automated vehicles.” The surveyed documents were not limited to journal articles but also included conference papers, technical reports, and abstracts of active and recent projects. For WZ-publications, the average number of annual publications experienced a remarkable climb in the 2000-2009 period as compared to the 1990’s, i.e., 179.9 versus 97.5 publications/year. WZ-publications continued to grow moderately in the last decade reaching 196.2 publications/year in the 2010-2019 period. On the other hand, only a handful of CAV-publications were noticed, if any, prior to 2010 after which the number of CAV-publications has been growing rapidly and even surpassing WZ-publications. However, when using the term “connected and automated vehicles at work zones,” only 12 records were retrieved, and these were in the period 2015-to-2019. Thus, deploying CAV systems at WZs is still at its early emergence and yet a plenty of publications are forthcoming as a natural consequence of the increasing trend of both WZ- and CAV-publications. ehman and Farooq 3
Notes: numbers were based on TRID database as of October 28, 2020, the covered period is from 1990-2019
Figure 1. Evolution of CAV and WZ publications in the last three decades
The previous discussion highlights two main points; first, the high importance of integrating or deploying CAVs at WZs, and second, the fact that this field of research is still at its early emergence. There is therefore a need for a paper that establishes a framework for this research direction and paves the way forward. This paper responds to this motivation by: (i) investigating how CAVs and WZs will interact and discussing the resulting key concepts, opportunities, and challenges, (ii) reviewing two categories of the literature, i.e., studies that particularly explored CAV concepts at WZs albeit only few of such efforts already exist and studies that have relevant and transferable concepts which can be applied at WZs, and (iii) presenting a research agenda supported by specific research needs.
The aim is to provide a bird’s eye view along with supporting details in order to map the deployment of CAVs at WZs and to benefit practitioners, planners, transportation agencies, and prospective researchers who are interested to understand the future of WZs. The paper first focuses on discussing how CAVs can be deployed to enhance WZ traffic operations and performance and the underlying mobility and safety concepts. Infrastructure considerations are then presented including traffic control devices and communication techniques. Subsequently, a research framework is outlined and research needs are listed. A conclusion summary is finally provided. N u m b e r o f P ub li ca ti on s average = 97.5 average = 179.9 average = 196.2 Year ehman and Farooq 4 TRAFFIC OPERATIONS AND PERFORMANCE
Automation is different than connectivity, vehicle’s functions may rely on one of them or both. However, this paper generally uses the abbreviation CAVs combining automation and connectivity without detailing the level and contribution size of each when discussing every concept. Nevertheless, in the following discussed concepts, the title of each indicates in parentheses whether the concept relies on connectivity, automation, or both technologies. This is indicated from a high-level perspective, so if a function is mainly a “connectivity-based” this does not mean it cannot also benefit from automation and vice versa. Most CAV functions are better to be performed or optimized with the presence of both automation and connectivity.
Mobility Concepts
Mobility at WZs refers to traffic operational measures such as capacity, demand, level-of-service, delay, and travel time. The objective is to optimize WZ capacity and also reduce or at least manage the upstream demand so the traffic can traverse the WZ smoothly and quickly. In the following sections, the discussed CAV mobility concepts were ordered such that the basic or universal concepts come first and also those concepts that rely on others follow them.
Tightened Following Gap (Automation- and Connectivity-Based)
CAVs are expected to shorten the following gap allowing to better utilize the highway. Nowakowski et al. (5) indicated that CAVs supported by V2V communication can reduce the mean following time of human-driven vehicles from 1.40 sec to 0.60 sec. A shorter following gap means higher capacity which is vitally needed at congested WZs. Most adaptive cruise control (ACC) systems follow a “speed-based gap” whereby the distance between vehicles is proportional to their speed plus a fixed standstill distance (6). This is especially important at WZ, where the traffic speed fluctuates widely. The tightened gap between CAVs is mainly based on their ability to accelerate and decelerate assertively and promptly. They use connectivity to obtain data of high-accuracy and low-latency of the lead vehicle motion and use automation to eliminate the imperfection in driver’s perception-reaction [7, 8].
CAV Clustering (Mainly Connectivity-Based)
If CAVs are distributed randomly in a mixed traffic, the chance of CAV-coupling will only depend on the market penetration rate (MPR). More CAV-generated short-gaps can be realized if CAVs can deliberately follow each other. CAV clustering instructs equipped-vehicles, that are travelling in close proximity, to form platoons and strings. To do so, CAVs may have to accelerate, decelerate, and change lanes. They have also to be provided with accurate data about the motion of nearby equipped-vehicles including their travel lane. Associating nearby CAVs with their exact travel lane can be challenging at WZs having complex and varying lane configurations. This warrants the use of special and smart traffic control devices as discussed later in Sections 3.2 - 3.4. Shladover et al. [6] defined “Global Clustering” when vehicles sharing the same destination meet early in the network to form platoons, the authors however considered this as challenging because it requires wide communication coverage and waiting areas for vehicles to meet. Although clustering is generally beneficial for congested traffic, it may result in adverse impacts on some lanes near the WZ taper. Figure 2 illustrates some possible impacts of clustering by different lanes. Clustering on the closed lane creates long platoons that may not be able to find sufficiently wide merging gaps on the adjacent lane (Figure 2a). If traffic demand is persistently large on the adjacent open lanes, dissolution of the stranded clusters may become inevitable. Additionally, if the rate of clustered vehicles is excessively high on the first open lane (Figure 2b), long and dense platoons will be passing repeatedly on that lane. Consequently, vehicles on the closed lane will not find merging gaps easily and they may have to endure excessive queuing delay. This problem can especially impact human-driven vehicles which cannot communicate with CAVs and ask for cooperation. Alternatively, clustering may be better limited to the far open lanes at high traffic demand conditions (Figure 2c). This insures easier ehman and Farooq 5 lane change maneuvers from the closed lane to the first open lane for both the human-driven vehicles and the non-clustered CAVs. Another advanced option is to assign a dedicated lane for CAVs (Figures 2d and 2e) which is further discussed in the following concept (Section 2.1.3). Permitting or prohibiting clustering on specific lanes and how this can be influenced by WZ lane configurations (2-to-1, 3-to-1, 4-to-1, 4-to-2, etc), traffic demand level, MPR, maximum platoon (cluster) length, and the distance remaining between the approaching clusters and the WZ taper are critical topics that merit a detailed investigation. ehman and Farooq 6 (a) Clusterin on the closed lane (b) Clustering on the first open lane (c) Clustering on the far open lanes (d) Clusterin on a dedicated CAV-lane with right-side closure (e) A dedicated CAV-lane with left-side closure
Notes: distances are not to scale and for illustration purpose only, maximum platoon length is assumed to be four vehicles
Figure 2. CAV clustering by different lanes in the WZ vicinity
Closed lane First open lane Second open lane Third open lane CAV-Lane CAV-Lane CAV-Lane CAV-Lane CAV-Lane Gaps are short for a full-cluster merge
Queued vehicles waiting for merging gaps
Queued clusters and vehicles waiting for merging gaps
Clusters reduce the frequency of wide merging gaps on the first open lane
CAV-platoon (or cluster) Platoon leader Platoon member Human-driven vehicle or non-clustered CAV Travel direction
Legend ehman and Farooq 7
Dedicated CAV Lane (Automation- and Connectivity-Based)
An advanced level of clustering CAVs is to assign them a dedicated lane, this lane however can still accept human-driven vehicles if the MPR is not high enough. A dedicated lane would mainly improve traffic efficiency and increase capacity by utilizing the smart car-following between CAVs while also promising for cluster-based safety and environmental benefits. The capacity of a dedicated CAV-lane can reach 3400 veh/hr/ln at 90% MPR [9]. Moreover, CAV-lane allows for more stable communication density which makes the communication more reliable with low chance of packet drop [9]. A minimum of 20% MPR in the network and a four-lane basic section are needed to justify a dedicated CAV-lane [10, 11]. A dedicated CAV-lane should be typically assigned on the furthest left lane in order to avoid conflicting with ramp junctions. This perfectly suits a right-side closure (Figure 2d). If the closure is on the left-side, a buffering taper is needed to maintain the CAV-lane continuity and shift the closure to the other mixed-traffic lanes (Figure 2e).
Synchronized Acceleration/Deceleration (Automation- and Connectivity-Based)
When human-driven vehicles discharge from a queue, vehicles in front cannot move simultaneously and they have to accelerate and discharge one after the other resulting in a start-up lost time. The V2V-powered following behaviour allows a platoon of vehicles to accelerate synchronously in a train or semi-train fashion eliminating or reducing the start-up lost time. With V2V communication, the fourth vehicle can be notified of the lead vehicle action within 400 to 800 ms if data are propagated pair-wise, or within only 100 ms if data are propagated platoon-wise upstream irrelevant to the line of sight [6]. The synchronized acceleration/deceleration can fit well in congested WZs where vehicles discharge from queued traffic. With a traffic signal, a V2I communication between queued vehicles and the signal can make the movement of train of vehicles perfectly synchronized with no loss in the green phase time. It is worth noting that traffic signals are not only suitable at arterial WZs but may also be deployed at freeway WZs as supported by recent studies. Ramadan et al. [12] and Tympakianaki et al. [13] evaluated the use of traffic signal on the freeway mainline immediately upstream the WZ in order to reduce lane change friction between vehicles and to provide an organized queue discharge discipline. The studies indicated favourable safety and mobility outcomes during peak-hour conditions.
Lane Merge Control (Automation- and Connectivity-Based)
Lane merge control at WZs is a well-established research subject and many merge strategies have been proposed and evaluated. Figure 3 demonstrates three different merge control strategies at WZs. In the late-merge (Figure 3a), drivers are advised to stay on their lane, they are then instructed to follow a zipper-merge control just immediately before the WZ taper. This approach can be disadvantageous to CAV operations since it results in splitting CAV clusters that were formed upstream. In the early-merge (Figure 3b), drivers are advised to change lane early upstream to avoid last-moment aggressive merging. This can protect CAV clusters especially if “CAV cooperative lane change” is deployed upstream to enhance the clustering. However, early-merge control does not perform well at congested sites, it shifts the queues further upstream, and in extreme cases, the early-merge will almost operate like a late-merge system with a merging point shifted upstream. With a traffic signal control (Figure 3c), vehicles occupy all lanes and follow a phasing scheme. A full-traffic-cycle instead of one-car-per-green is deployed and each lane is provided with a distinct traffic light so the lanes do not discharge traffic simultaneously. This approach protects CAV clusters and allows for a platoon-wide synchronized acceleration using the V2I communication between CAVs and the signal controller. The phasing can be fixed allowing for (n) vehicles to discharge per green or can be actuated based on the available clusters, i.e., the green time may be slightly extended or shortened to allow clustered vehicles to discharge at once. A detailed comparison between the three merge control strategies is recommended as a future research. ehman and Farooq 8 (a) Late-merge control system (zipper discipline) (b) Early-merge control system (front vehicles in the open lane leave first) (c) Traffic signal control system (per-lane phasing alternation discipline)
Note: the numbers refer to the order of the discharged vehicles
Figure 3. Different lane merge control systems
Cooperative Lane Change (Mainly Connectivity-Based)
The cooperative lane change fits well with closure-based WZ areas because a plenty of vehicles need explicitly to merge to the open lanes. The most common cooperative lane change concept typically involves a pair of vehicles as shown in Figure 4a. The first vehicle (Vehicle A) signals to nearby vehicles explicitly that it needs to change its lane. The second vehicle (Vehicle B), which is travelling on the target lane immediately upstream the first vehicle, will recognize the signal and then decelerate to allow for a smooth and reliable lane change. Vehicle B on the target lane may also collaborate by switching to the next adjacent lane so it creates adequate merging gap for vehicle A (Figure 4b). When traffic demand increases, the decision of which vehicle needs to collaborate with which vehicle becomes more complex as shown in Figure 4c. If vehicles A1 and A2 signal their need to change lane, vehicle B1 may decide that it is too late to cooperate and it therefore speeds up and moves forward. Then, vehicle B2 is left with a complex decision, i.e., if it decided to cooperate, it needs to perceive the projected trajectories and speeds of both vehicles A1 and A2 and decide accordingly which one is more feasible to collaborate with. Cooperative lane change for such complex traffic flow situations can be addressed by incorporating sophisticated algorithms [14]. Alternatively, Ren et al. [15] proposed a strategy whereby vehicles on the open lane are instructed to maintain a minimum distance from each other (i.e., equal to two times the safe distance) early upstream of the WZ. When the WZ taper becomes close, vehicles on the closed lane can subsequently merge easily to the open lane as in Figure 4d. The applicability of this approach however requires 100% MPR. V2V communication should be used for all cooperative concepts because V2I communication may not transmit data with sufficient latency and accuracy, the V2V communication delay however must still be carefully considered [16]. Research is recommended to elaborate more on developing cooperative lane change algorithms and strategies that consider different MPRs, urgency of the lane change request, Blocked zone – drivers cannot pass and must merge early Travel direction ehman and Farooq 9 the distance remaining to the WZ taper, communication latency, WZ lane configuration, traffic demand level, and the possibility of cooperating with human-driven vehicles. (a) Deceleration-based cooperation (b) Lane-change-based cooperation (c) Complex cooperation situation in a dense traffic flow (d) Global-level cooperation
Figure 4. Cooperative Lane Change Scenarios
Smart Lane Flow Distribution (Mainly Connectivity-Based)
Achieving a balanced flow distribution over the traffic lanes will optimize capacity and enhance safety. A freeway section usually has larger flow on the fastest left-side lanes. This generates a heavy demand for lane changes at a WZ with left-side closure. CAV-powered smart lane flow distribution depends on early and gradual lane assignment strategies whereby the system randomly selects a group of drivers to change lane, then another group and so forth until a desired lane balance is achieved. Drivers who still have not changed lane from the preceding group are given priority, the number of simultaneous lane changes must be limited. Schakel et al. [7] used V2I communication to advise drivers to change lanes aiming to achieve an optimal lane flow distribution. The system was applied at several highway configurations including a lane-drop case, it delayed the onset of breakdown and shortened the overall travel time by 49%.
Early Rerouting (Mainly Connectivity-Based)
Rerouting schemes are common traffic management applications at WZs, they advise drivers of available alternate routes. While this initially reduces average trip travel time, it also results in lowering traffic demand at the WZ. CAV-based rerouting at WZs can be smarter in different ways. First, the wide communication coverage of CAV systems allows drivers to alter their routes early at the onset of their trips based on their origin-destination, they do not have to wait until seeing advisory signs upstream the WZ area. Second, a multitude of rerouting options can be provided. Third, traffic conditions on the alternate routes will be updated more frequently and transmitted with low latency allowing for timelier decisions.
A B AB A1 B2B1 A2
A1B1 B2 B3 B4A2 A3
Travel direction ehman and Farooq 10 Figure 5 outlines the spatial distribution of the previously discussed mobility functions and concepts at WZs. The WZ area is subdivided into five sub-areas each representing specific traffic conditions and driver’s tasks. The range of each function denotes where it is mostly needed and effective. For example, adaptive cruise control is very effective at the queuing area until discharging from the termination area, it can still however be used early upstream if desired by the driver.
Figure 5. Spatial distribution of CAV mobility functions at WZ areas
WZ Termination Area WZ Activity Area WZ Queuing Area WZ Approach Area Further Upstream Tightened Following Gap (Adaptive Cruise Control - automation levels 2 and 3 and) Early Lane Change (Merge) Advisory Rerouting Advisories (early in the network) Rerouting Advisories (immediately upstream) CAV Clustering (on the far open lanes) Synchronized Acceleration and Deceleration Cooperative Lane Change Smart Lane Flow Distribution CAV Dedicated Lane (≥ 4-lane section) Lane Merge Control ehman and Farooq 11
Safety Concepts
Safety at WZs refers herein to reducing or eliminating the chance of accidents by relying on the vehicles, the WZ settings, the drivers, or all together. Vehicle automation is discussed first because of its universal role in eliminating drivers’ errors. Subsequently, connectivity-based concepts specific to WZ vicinities are introduced.
Vehicle Automation (Mainly Automation-Based)
Many studies indicated that human errors are the major causes of WZ crashes [17-19]. Chambless et al. [17] estimated that human factors contribute to around 82.7% of all WZ crashes. The high proportion of driver-related crashes makes vehicle automation a promising and vital countermeasure. When automation carries out the driving tasks, driver’s errors and imperfections will be eliminated. In an optimistic estimation, all crashes caused by driver-related reasons can be avoided. Automation benefits however will be restricted by some factors, e.g., MPR, automation level, supplying the WZ area with automation-needed infrastructure, and the automation malfunction or error rates if likely. However, the literature suggests that average levels of automation (Levels 1 to 3) will penetrate the market heavily in the short-run, these levels allow for adaptive cruise control which well suits WZ conditions like jam and stop-and-go conditions.
Speed Harmonization (Mainly Connectivity-Based)
All vehicles have to frequently change their speed when traversing a WZ depending if they are approaching the WZ, traversing the work activity area, or discharging downstream. When congestion prevails, the stop-and-go regime adds extra speed oscillation. Speed variability constitutes a major safety hazard. Speed harmonization is a countermeasure technique that adjusts the advisory speed dynamically based on real-time data. A simple speed harmonizer can be a V2I system that measures the speed within the WZ using a roadside unit (RSU) and transmits advisory messages to approaching drivers. For more advanced speed harmonizers, the highway can be subdivided into several segments each having its own RSU that measures the local speed and advises the equipped-vehicles accordingly. Recent researches [e.g., 20, 21] showed that CAV-powered speed harmonization can be useful even with low MPRs.
Crash Warnings (Mainly Connectivity-Based)
The literature almost agreed that rear-end collision is the most predominant type of WZ crashes and that the other common types include sideswipe, angle, and hit-fixed-object collisions [19, 22, 23]. These crashes also follow important spatial trends [22]. Rear-end collision is the most prevailing type everywhere along the WZ area; however, the frequency of sideswipe collision critically increases in the transition area, and the frequencies of the angle and hit-fixed-object collisions critically increase in the work activity area. The prevalence of common crash types at WZs and their spatial trends suggest establishing specific CAV-powered crash warnings. First: Rear-End Warning A V2V communication should be established so the following vehicle can receive from the leading vehicle its position, speed, and acceleration. When a crash becomes imminent, a standard warning message can be triggered and transmitted to the following (striking) vehicle . Second: Sideswipe and Angle Collision Warning Warning for sideswipe and angle collisions needs information on all vehicles travelling in adjacent lanes. This warrants accurate V2V communication that can detect the exact lane of nearby vehicles while also measuring their motion features. Third: Work-Activity Area Encroachment Warning Workers can have wearable connectivity devices that track their position and activity. Smart wearable devices are already in use in the construction industry and they include, for example, smart vests, watches, wrist bands, helmets, and shoes [e.g., 24, 25]. Similarly, traffic control devices (e.g., cones and barrels) and construction equipment can be equipped with smart devices that disclose their identity and location. ehman and Farooq 12 Using V2I (RSU) or V2X (direct communication), approaching vehicles can be alarmed when becoming too close to workers, devices, or equipment. The workers can also be alarmed by receiving audible, visual, haptic, and multi-modal alarms in their wearable devices. All crash warnings are primarily dependent on V2V and V2X connectivity; however, vehicle automation would also enhance the perception-reaction time to any warning message. In the absence of V2V communications, a warning message can still be transmitted to the striking vehicle with only V2I communication, but the struck vehicle should be adequately equipped and connected to the RSU so it can disclose its position and motion characteristics. Nevertheless, V2I communication will be restricted by the RSU range and the limited latency may adversely impact the timeliness of the warning message.
WZ Presence and Queue Warning (Mainly Connectivity-Based)
Warning drivers of a WZ or end-of-queue ahead can be done using unsophisticated communication. V2I is needed but V2V can increase coverage range when only one RSU is available. Like mobility functions, Figure 6 outlines the spatial distribution of safety functions and concepts at WZs.
Figure 6. Spatial distribution of CAV safety functions at WZ areas
WZ Termination Area WZ Activity Area WZ Queuing Area WZ Approach Area Further Upstream WZ Activity Area Encroachment Warning Presence of WZ Ahead Warning Vehicle Automation Assistance (level is based on comfort) End-of-Queue Warning Vehicle Automation Assistance (Levels 2 & 3 are effective) Speed Harmony Rear-End Collision Warning Sideswipe Collision Warning ehman and Farooq 13
Demand of Automation and Connectivity at WZs
Figure 7 outlines the demand level of automation and connectivity in the WZ vicinity which is a high-level analysis inferred from the previously discussed concepts. Generally, the need for automation increases as vehicles move downstream whereas the need for connectivity is highest upstream. In the WZ approach area and further upstream, vehicles’ functions are mainly connectivity-based (e.g., rerouting, early merging, lane change cooperation, receiving WZ and end-of-queue warnings, forming ideal flow distribution). The congestion level is usually lower upstream with no queuing, vehicles so do not heavily need automation. At both the WZ queuing and activity areas, both connectivity (hazard detection, collision warnings, platooning) and automation (tight coupling, stop-and-go navigation, eliminating drivers’ errors and imperfections) are heavily needed. Finally, downstream the WZ area, automation is needed to discharge the queues effectively by controlling the acceleration and the gaps, the demand of connectivity decreases but it is moderately needed to enhance vehicles clustering and their synchronized acceleration.
Figure 7. Level of deployment of automation and connectivity at WZs
WZ Activity Area WZ Queuing Area WZ Approach Area Further Upstream WZ Termination Area High
WZ-related vehicle connectivity deployment level
Highest High Moderate Low
WZ-related vehicle automation deployment level
Moderate High High Highest Highest Overall increase of automation need Overall increase of connectivity need ehman and Farooq 14
CAV Mobility and Safety at WZs: Existing Literature
Only few studies have elaborated on the deployment of CAV systems at WZs, Table 1 summarizes the main efforts found in the literature focusing on safety and mobility aspects. Interestingly, almost all the studies relied on simulation tools in the absence of considerable CAV presence in the field currently. Studies should however consider field testing as well in the short-term, only one study relied upon field testbed. The cited studies overall indicated promisingly positive results. These included capacity enhancement, travel time and delay reduction, smoother and safer merges, harmonized speed, enhanced driver’s awareness, and reduced chance of rear-end collision and WZ area encroachment. At this stage when only few efforts exist, conclusive quantitative results cannot be drawn reliably but some indicative or general lessons can be learned. Overall, the reported size of CAV benefits at WZs appears to be large, but this was critically sensitive to the MPR and the level of traffic flow at WZ. Zou et al. [26], for example, found that CAV MPR of 34.1%, 62.25%, and 100% achieved average travel time reductions of 25%, 50%, and 90% respectively at a congested 2-to-1 WZ configuration. Abdulsattar et al. [27] tested the impact of connected vehicles at several traffic demand levels. For example, for a demand level of 3000 vph and MPR of 75% of connected vehicles, the mean travel time was reduced by 40% and capacity increased by 65% at 2-to-1 WZ. Only at high traffic demand levels the benefits of connected vehicles were significant. In addition, for each traffic flow level, there was a critical range of MPR where the benefits of connected vehicles became maximized and after which there would be no more feasible benefits by increasing the MPR. Genders et al. [30] evaluated the safety benefits resulting from rerouting connected vehicles away from WZs. The time-to-collision was used a safety surrogate measure. Moderate MPRs (<40%) enhanced the network safety because the improved driving behaviour overweighed travelling additional trip distances; however, the network safety degraded for large MPRs (>40%) because substantially longer trip distances were added to the network and this increased the exposure to safety hazards and made the enhanced driving behaviour less effective. Abdulsattar et al. [31] examined the impact of connectivity on reducing rear-end collisions and indicated that the first entry-level of 10% MPR resulted in significant safety improvement. For example, at medium and high traffic flow rates, the first 10% MPR decreased the critical time-to-collision by 50%, i.e., the probability of rear-end collision was reduced by 50% at 2-to-1 WZ. Beyond the first 10% MPR, higher traffic flow rates required higher MPRs to show further safety improvements. Mollenhauer et al. [33] described and field-tested a WZ hazard detection system whereby all WZ actors (i.e., workers, vehicles, equipment, and cones) where tagged and connected. The system was trained, using machine-learning techniques, to recognize and classify the activities and movements of the workers (i.e., jackhammering, walking, rolling, guiding, or random). The trajectories of CVs and equipment were also tracked by the system that sends accordingly warnings to vehicles or workers when a hazard becomes imminent. Hazard detection algorithm was set to be sensitive to the type of undergoing worker’s activity, worker’s proximity to the WZ border, prediction of actors’ movements, highway geometry (curved or straight), WZ barrier type, blind spots, and the vehicle stopping distance (i.e., speed, pavement friction, and reaction time). The system further considers sending a specific escape direction to workers when CAVs are only instructed to brake. More studies and findings can be learned from Table 1. The existing literature only explored a limited number of CAV concepts. For example, mobility studies focused on deploying smart car following and collaborative lane change, and safety studies examined the provision of in-vehicle warnings, rerouting, and speed harmonization. More efforts are needed and recommended to explore all the mobility and safety concepts discussed in the previous sections including, for example, clustering techniques, synchronized acceleration, smart lane flow distribution, dedicated CAV lane, and the different levels of vehicle automation. Combining several CAV concepts at one WZ site may magnify the resulting benefits and this also deserves further evaluation. Nevertheless, attention must be paid to the driving behaviour used or assumed in the analysis. The cited mobility studies were generally interested in assertive driving behaviour and shorter following gaps whereas safety studies tended to assume conservative following behaviour and longer gaps. This indicates that CAV benefits are sensitive to the assumed driving behaviour and achieving both sizeable mobility ehman and Farooq 15 and safety benefits at once merits further investigation, e.g., by exploring the reliability of automation and sensing of the leading vehicles in a mixed traffic.
Table 1. Literature Review Summary for CAV Mobility and Safety Functions at WZs
Authors Main Theme Study Scope Analysis Tool Main Findings
Zou et al. [26]
Mobility Explored how CAVs would impact several WZ traffic operational measures (travel time, emissions, speed harmony). Cooperative Cellular Automata Model (CCAM) simulation
CAV deployment reduced travel time and emissions and increased speed harmony. Larger MPRs enhanced these benefits and reduced the stochasticity in the measured variables.
Abdulsattar et al. [27]
Mobility Explored the impact of vehicle connectivity (V2I and V2V) on WZ traffic speed, capacity, travel time and its variability. Agent-based modelling and simulation (ABMS)
The benefits of CVs (larger capacity, higher speed, and lower mean travel time) become noticeable and significant only at high traffic flow rates.
Ren et al. [15]
Mobility Proposed and evaluated a collaborative merge control technique at WZs.
Vissim microscopic simulation software
The cooperative merge control technique proved to outperform other traditional merge control systems. For a mixed traffic, the applicability of the technique needs further investigation.
Weng et al. [28]
Mobility Proposed a merging assistance system at WZs that guides safe merging. Used vehicle trajectory data and non-paramteric method called “CART” The most influencing factors on merging behaviour were the merging time elapsed, remaining distance to WZ, speed, and the time-to-collision between the merging vehicle and neighbouring vehicles.
Ramezani et al. [29]
Mobility and Safety Developed and evaluated a connected-vehicles-based speed harmonizer methodology at WZs. LOQO solver for large-scale nonlinear optimization The delay was reduced by 13% when the MPR was 80% or higher. During 40-minute operation period and 100% MPR, the congestion period was reduced by 26.4%.
Genders et al. [30]
Safety & Driving Behaviour Evaluated the impact of providing connected vehicles with warning messages of the WZ presence on the network time-to-collision (TTC). Paramics microscopic simulation software Early dynamic rerouting enhanced driving behaviour but this also increased average trip distance which in turn enlarged the exposure to safety hazards especially with high MPR.
Abdulsattar et al. [31]
Safety Evaluated the impact of deploying vehicle connectivity on the probability of rear-end crashes at WZs. Agent-based modelling and simulation (ABMS) The first entry-level of 10% MPR resulted in substantial safety improvement. Beyond the first 10% MPR, higher traffic flow rates required higher MPRs to show further safety improvements.
Qiao et al. [32]
Safety & Driving Behaviour Evaluated smartphone-based rear-end collision warning messages at WZs. Driving simulator environment Voice and sound warnings were effective at reducing speeds and prolonging headways. Visual warning deteriorated driving behaviour and generated visual distraction.
Mollenhauer et al. [33]
Safety Designed and evaluated a WZ hazard detection system where all WZ actors (i.e., workers, equipment, vehicles, and cones) were connected Field testbed The demonstration experiments successfully verified real-time data acquisition and the transmission of safety messages. Tagged objects were positioned within an accuracy of approximately 10 cm within a system range of 80 m. ehman and Farooq 16 TRAFFIC CONTROL DEVICES
Traffic control devices (TCDs) are indispensable at WZs where the roadway geometry usually changes suddenly resulting in a complex configuration. TCDs at WZs differ from other standard roadway sections because they usually include signs with more complex contents, temporary devices (cones, barrels, markings), flaggers, and others. CAVs need to perceive these TCDs reliably and promptly so they can traverse the WZ efficiently and safely. The following sections provide an overview of several concepts that facilitate the interaction between CAVs and several types of TCDs at WZs.
Smart Sign Detection
At high automation levels, vehicles should autonomously recognize and react to the road signs. Even when drivers are in control of the vehicle, vehicle-based sign detection can assist by providing prompt in-vehicle visual or voice notification of the sign.
Camera-Based Sign Detection
Camera-based traffic sign detection is a well-established research area that uses sophisticated and computerized vision techniques. Under these systems, the vehicles use camera recordings of the roadway ahead. The video images are then analysed using computerized techniques (e.g., neural network [34, 35], deep learning [36], support vector machine [37], random forest [38]) that classify and recognize the sign based on an inventory of predefined standard signs. Lee et al. [39] tested a Kernel vision-based sign detection system at WZs. The system reduced false detection rates as compared to other vision systems, but the authors stated that different road and illumination conditions should be further tested. According to the literature [e.g., 40, 41], challenges of the camera-based sign detection also include: (i) the lack of a standard sign database that can be generalized everywhere, and (ii) the likely that the system may detect a sign that does not belong to the vehicle route (e.g., signs on intersecting roads). Furthermore, sign detection systems usually have a limited database capacity and WZ-specific signs may not be given priority since they treat temporary conditions.
Lidar-Based Sign Detection
Light detection and ranging (LIDAR) is a remote sensing technology that illuminates target objects with laser light and the reflected light is then detected and analyzed by sensors. The sensors can measure distances and create coloured 3D images which enable Lidar-equipped vehicles to locate obstacles and recognize signs. LIDAR detection is also a computerized vision-based method, but it infers the images using the reflective intensity data and not video recordings. Many recent efforts indicated promising results for using LIDAR-based sign detection methods [e.g., 42-45]. These studies also emphasized that the reliance of LIDAR on the sign retro-reflectivity features provides more detection robustness as compared with camera-based detection methods which can be impacted by light conditions, image quality, and occlusions. This allows to detect more types of signs and to manage a larger inventory of traffic signs. LIDAR detection also provides robust 3D information on the surrounding setting. However, one challenge to LIDAR is that laser reflectivity can be impacted by the sign pose. Some researchers recommended a hybrid system that relies on the fusion of complementary data from two sensors, i.e., a camera and LIDAR, in order to further improve the robustness of the sign detection algorithm [e.g., 46].
Communication-Based Sign Detection
A V2X communication can be established between a vehicle and a sign equipped with a tag that transmits a message defining the sign identity. Garcia-Garrido et al. [41] proposed a radio frequency V2X communication system that uses wireless sensors installed on the traffic sign post, the sensors transmit a message to an on-board unit (OBU) including information about the sign type, its position, and the road name. The sign was able to communicate to all equipped-vehicles within a 500 m coverage, the study also indicated the possibility to widen the coverage. Qiao et al. [47] similarly developed and successfully tested a radio frequency V2X sign detection system at two pilot WZ test routes, the sign message prompts were activated based on the stopping sight distance upstream the sign. Liao et al. [48] also successfully tested a Bluetooth-based V2X between drivers’ smartphones and Bluetooth-tagged signs. ehman and Farooq 17 The communication-based sign detection appears more favourable for CAV operations at WZs, it resolves the limitations of the vision-based detection, i.e., it discards irrelevant signs, it eliminates the chance for false detection, and it customizes sign identification (message content and time) according to each site characteristics.
Smart Lane Identification
An important navigation challenge for CAVs at WZs is the sudden change in lane configuration and the resulting complex geometry. CAVs cannot solely rely on digital maps, these may not be updated promptly. This is a special problem for short-term WZs that may last only few days or even hours. There is therefore a need to use lane identification system that can communicate with CAVs reliably and seamlessly. An option is to use unique lane markings (i.e., with specific shape, dimension, and color) that can be easily recognized by computerized camera-based detection systems. In addition, LIDAR-based detection has been reported as a promising technique to detect and track lane markings [e.g., 49-51] that is less sensitive to shadow, sunlight, or night as compared to camera-based detection. The LIDAR cloud images do not necessarily rely solely upon the painting retroreflectivity. However, the retroreflectivity of the marking paint can make its laser reflectivity value larger than the surrounding and hence more discriminable. A more advanced option is to use radio frequency identification chips either embedded in the pavement or placed as lane dividers [6]. It is important to note that these smart lane marking techniques can also be deployed to locate other nearby CAVs making those functions that rely on the positioning of other vehicles more efficient.
Smart Cones
Smart cones refer to wirelessly connected cones that are supplied with sensors that are intended primarily to detect errant vehicles and sometimes pedestrians trespassing the WZ area. The system sends voice and visual alarms to workers on site if a hazard is detected and it also sends a request to remove the debris and reinstall cones when a vehicle hits a WZ taper. The literature review conducted herein did not reveal scholarly papers that comprehensively analysed smart cones albeit they have been applied widely as evidenced by the presence of many manufacturers [e.g., 52, 53]. Nevertheless, the functionality of these devices can be easily extended by establishing a communication with approaching CAVs which can subsequently easily discern the cones and identify the WZ taper boundary. Transitioning from traditional to smart cones may therefore become more warranted with the forthcoming penetration of CAVs. To be noted that traditional cones can still be detected using camera-based image detection algorithms as suggested by Wang et al. [54]. However, the connected cones allow for more detection reliability and accuracy.
Robotic Traffic Control Devices (RTCDs)
RTCDs are intelligent devices that can move autonomously while being connected to a central robot base. RTCDs can mainly include barrels, cones, barricades, and sign bases. Changing the positions of the RTCDs dynamically, flexibly, and remotely based on the work activities can reduce the size and duration of WZs and reduce workers’ exposure to traffic. RTCDs must move reliably, they may otherwise generate hazards. Shen et al. [55] developed and tested robotic safety barrels which move autonomously at WZs using a central station which localizes the barrels and communicates the planned path for each. The system was tested in realistic highway environments and the maximum final positioning error for all robots was 11 cm exceeding human accuracy of barrel placement. The system was intended to operate with human-driven vehicles. However, CAVs can easily communicate with RTCDs when approaching the WZ area through V2I or V2X channels. ehman and Farooq 18
Automated Flaggers
Automated flagger assistance devices (AFADs) have been already applied in some states and countries at WZs [56-58] allowing to operate an automated device (e.g., Stop-Slow signs, Red-Amber lights) from a communication base away from the roadway. The deployment of these devices initially aimed at enhancing safety by reducing human exposure to traffic and reducing labour cost if applied for long durations. Some challenges of AFAD deployment were noticed among researchers, e.g., drivers had difficulty to understand the system and both the approaching speed and the deceleration rate increased raising the risk of rear-end collisions [56-58]. However, by establishing a V2I communication between CAVs and AFADs, the deployment of AFADs at WZs can be greatly revitalized. CAVs cannot communicate easily with human flaggers and therefore they need AFADs especially under high automation levels. Moreover, CAVs would receive standard and prompt messages that allow for clearer understanding of the system and smoother deceleration. ehman and Farooq 19 COMMUNICATION
This section discusses the communication technological aspects between WZs and CAVs. First, the main potential communication technologies at WZs are introduced. Communication types and needs are presented afterwards. Finally, the available literature that investigated CAV communication at WZs is summarized.
Communication Technologies
Two important parameters of communication technologies for transportation applications are latency (time between when the information becomes available for broadcast and when it is received) and range (distance between two communicating units needed to have an efficient communication). Low latency and long range are desired but unfortunately they cannot coexist, technologies that offer low latency suffer from narrow range and vice versa. Practitioners must choose the communication technology based on the needed applications.
Dedicated Short-Range Communication (DSRC)
Many studies considered DSRC as the most promising communication technology for CAV applications in the near term [59-63], this is valid for V2I, V2V and hybrid communications. DSRC has many advantages that well fit transportation-oriented applications including high reliability, low latency, interoperability, and security. The most fitting applications to DSRC are safety and crash warnings. The literature agreed that crash warning applications require latency as low as 100 ms [e.g., 64], this makes most non-DSRC communication technologies inefficient for WZ crash warnings. Vehicle clustering and tight coupling at WZ also need low latency and DSRC capabilities. However, one main disadvantage of DSRC is the limited range and scalability.
Cellular Communication
Capabilities of early cellular technologies are opposite to those of DSRC, i.e., they offer wider range and larger packet data but with higher latency. Xu et al. [65] compared 4G-LTE cellular technology with DSRC and found that DSRC clearly outperformed 4G-LTE for safety application; however, due to its wider availability and higher throughput, the 4G-LTE was recommended for application that require information transmission and file download. However, the rapid evolution of cell-phone consumer market has made this technology more deployable. Recent researches indicated that the newly introduced 5G-based communication technologies have small latency and can be deployed efficiently for CAV communications and applications [e.g., 66-70]. The arrival of 5G-cellular communication has enhanced the cellular vehicle-to-everything communication concept (C-V2X) which comprises communication types beyond V2V and V2I, e.g., V2P (vehicle-to-pedestrian), V2D (vehicle-to-device), and V2N (vehicle-to-network). The 5G C-V2X appears so to be a possible competitor to DSRC for time-critical applications especially if it succeeds in combining low latency with wide coverage.
Satellite Communication
Global Positioning System (GPS) communication technologies offer massive coverage (hundreds of kms) but with very high latency (can reach 10-20 seconds). However, they can still be deployed in some mobility applications such as WZ presence warning, weather alerts, and alternate route advisory. Constellations of low earth orbit (LOE) satellites (e.g., SpaceX and One-Web) are growing rapidly and they may revolutionize satellite-based communications. These mega-constellations have a large number of satellites (hundreds or thousands) that are close to earth allowing for lower latency, precise positioning that may outperform the GPS, and broadband internet connectivity that can support various in-vehicle functions [71-74]. ehman and Farooq 20
Communication Types at WZs
Figure 8 illustrates the main communication types at WZs including V2I, V2V, and V2X. Table 2 also outlines the essential communication needs for all the aforementioned CAV functions and concepts pertaining to mobility, safety, and TCDs. This table was based on the essential functions of each application and the available literature. Nevertheless, exceptions may be observed for specific systems and the table may be further fine-tuned when more advanced technologies and more research findings become available. (a) Basic communication types (V2I and V2V) (b) Additional V2X communication types
Figure 8. Different CAV communication types at WZs
V2I V2I Connected Vehicles Human Driven Vehicles RSU Range
RSU
Communication with equipped barrels, cones, and workers Communication with equipped portable changeable message signs which can then post messages to human-driven vehicles Identifying lanes by detection of special lane markings or by communicating with embedded chips or sensors Communication with tagged signs Additional RSUs may be used upstream if needed (e.g., if MPR is low) ehman and Farooq 21
Table 2. Communication Types and Needs for CAV Functions at WZs
Function Essential Communication Needs Type Desired Features Example Technology
Tight-Coupling V2V Low latency, high reliability and accuracy DSRC, 5G Cellular* CAV Clustering V2V Low latency, high reliability and accuracy DSRC, 5G Cellular* Synchronized Acceleration V2V and can be enhanced with Hybrid V2I+V2V Low latency, high reliability and accuracy DSRC, 5G Cellular* Smart Lane Flow Distribution V2I Wide range Cellular, Satellite, and Scalable-DSRC Dedicated CAV Lane V2I Wide range Cellular, Satellite, and Scalable-DSRC Early Rerouting V2I Wide range Cellular, Satellite, and Scalable-DSRC Vehicle Automation V2V Low latency, high reliability and accuracy DSRC, 5G Cellular* Speed Harmonization V2I, V2V, or Hybrid V2I+V2V Wide range Cellular, Satellite, and Scalable-DSRC Work Zone Presence Warning V2I Wide range Cellular, Satellite, and Scalable-DSRC Rear-End Warning V2V Low latency, high reliability and accuracy DSRC, 5G Cellular* Sideswipe Warning V2V Low latency, high reliability and accuracy DSRC, 5G Cellular* Work-Activity Area Crash Warning V2X Low latency, high reliability and accuracy DSRC, 5G Cellular* Vision-Based Sign Detection (Camera-based, Lidar-based, or hybrid) vehicles are equipped with camera-based computerized systems or LIDAR sensors or both, communication is not needed Communication-Based Sign Detection V2X Low latency, high reliability and accuracy DSRC, 5G Cellular* Vision-Based Smart Lane Identification (Camera-based, Lidar-based, or hybrid) vehicles are equipped with camera-based systems or Lidar sensors or both, communication is not needed Communication-Based Smart Lane Identification V2X Low latency, high reliability and accuracy DSRC, 5G Cellular* Identification of Smart Cones and Robotic TCDs V2X Low latency, high reliability and accuracy DSRC, 5G Cellular* Automated Flaggers V2I Low latency, high reliability and accuracy DSRC, 5G Cellular* (*) needs further confirming research and testing ehman and Farooq 22
CAV Communication at WZs: Existing Literature
Table 3 summarizes the existing literature efforts that designed or tested CAV communication technologies at WZs. The Bluetooth and Ultra-Wide Band technologies provided limited range [48, 33] making them only applicable for short-range applications like sign detection. The cellular internet communication allowed for a network-wide coverage but with limited latency [77]; however, as stated previously the recent 5G cellular technology promises for low latency and deserves further testing at WZ sites. The one well researched technology at WZs that provided a trade-off between coverage and latency is the DSRC as explored by Maitipe et al. [62, 63], Zaman et al. [75], and Ibrahim et al. [76]. In these studies, approaching drivers were provided with information regarding the expected travel time and congestion location (queue length upstream the WZ). It is useful to summarize below the lessons learned from these studies which enhance the transferability and capability of DSRC systems at WZs.
Scalability of DSRC System
The V2I-DSRC system range can be increased by using several RSUs systematically spaced upstream the WZ. For V2V-supported DSRC systems, the communicated messages can reach any upstream distance as long as there are equipped-vehicles that continue to relay the message (i.e., sufficient MPR). To resolve for low MPR, vehicles on the opposite direction may also be incorporated in the rebroadcast channel [75].
Communication Load
In dense traffic, communication may create a broadcast storm if each vehicle receiving a message is relaying it forward. This can be important for WZ areas where dense traffic is common. A “Selective Relay” protocol was used to ensure that only one vehicle, selected in the farthest band distance from the last-originating vehicle, will relay the message [62]. The route can also be divided into concatenated rectangles and subrectangles, and only the farthest vehicle from the beginning of each subrectangular relays the message [75].
Communication Region
Communication should only be maintained on the WZ roadway, messages should not be exchanged with vehicles on parallel or intersecting roads. Maitipe et al. [62], established protocols to define an angular region around the WZ roadway and to check the road horizontal curvature, vehicles travelling outside the region were instructed to drop any received message immediately. Zaman et al. [75] used geographically-defined concatenated rectangles to predefine the hopping routes, the length of these rectangles decreases when the highway curvature becomes sharp and increases on tangent stretches. If an equipped-vehicle received a message outside the rectangles, i.e., on parallel or intersecting roads, then it will ignore the message. If desired, upstream on-ramps may be provided with similar predefined rectangles so they become integrated.
Communication Directionality
The messages should be propagated in one direction to avoid congestion and confusion. For messages originated from the RSU and propagated upstream, e.g., travel time estimation and end-of-queue warning, the receiving OBU only relay the message if it is farther from the RSU. Conversely, for messages originated from approaching vehicles and destining the RSU, e.g., vehicle’s location and speed, the receiving OBU only relay the message if it is closer to the RSU.
Required Market Penetration Rate
The efficiency of WZ communication systems strictly relies on having sufficient MPR. The RSU relies on passing equipped-vehicles to measure real-time information (e.g., travel time and end-of-queue location), it will therefore have to use unrefreshed and obsolete information if no new equipped-vehicles passed by. Additionally, the V2V back-and-forth message propagation will be halfway interrupted if no equipped-vehicle is found to receive and continue relaying the message. Ibrahim et al. [76] proposed a model, along with user-friendly charts, that estimates the MPR needed for the DSRC-based V2I and V2V ehman and Farooq 23 systems at WZs. The required MPR varies according to the desired information update frequency, DSRC range, and traffic flow rate. A higher MPR is needed if disseminated information are updated more frequently and if DSRC range and traffic flow are low. As an example, an MPR of 20% was needed for a 10-minute information update frequency, 250-meter average OBU range, and traffic flow of 1300 to 1800 vphpl.
Table 3. Literature Review of Field-Tested CAV Communication Technologies at WZs
Authors Communication System Range System Function Type Technology
Maitipe et al. [62]
V2I Dedicated Short-Range Communication (DSRC) The RSU range extended around 1 km (0.75 km upstream and 0.25 km downstream). The RSU was placed immediately before the transition area. Informing approaching vehicles of the estimated travel time and end-of-queue
Maitipe et al. [63]
V2I and V2V Dedicated Short-Range Communication (DSRC) The system can monitor congestion length of up to a few kilometers and the message broadcast coverage can reach few tens of kilometers. The range is strictly limited by V2V communication and available MPR. Informing approaching vehicles of the estimated travel time and end-of-queue
Zaman et al. [75]
V2I and V2V Dedicated Short-Range Communication (DSRC) The ranges of the RSU and the on-board unit (OBU) were 500 m and 250 m respectively. The range of the message propagation can be extended upstream as along as equipped vehicles are able to continue relaying the message. Informing approaching vehicles of the estimated travel time and end-of-queue
Ibrahim et al. [76]
V2I and V2V Dedicated Short-Range Communication (DSRC) The system deploys DSRC-equipped portable changeable message signs (PCMSs) that receive and display the same broadcasted information benefitting unequipped vehicles as well. The study also provided charts to estimate the required MPR for continuous and uninterrupted communication. Informing approaching vehicles of the estimated travel time and end-of-queue
Qiao. et al. [47]
V2X Radio Frequency Identification (RFID) The sign warning range was actuated based on the stopping sight distance. Two distances were used ahead of the signs, i.e., 250 and 305 ft upstream. The study did not discuss if a longer range is possible. Detection of tagged traffic signs
Azadi et al. [77]
V2I Cellular Internet Data The range is only limited by the cellular internet data access. The application does not work in dead zones or where the coverage is weak. General Smart Phone Applications (e.g., route planning, travel information, etc)
Liao et al. [48]
V2X Bluetooth Low Energy (BLE) Bluetooth tags were detected 125 m and 150 m upstream when travelling at 70 mph and 55 mph respectively. Detection of tagged traffic signs
Mollenhauer et al. [33]
V2I Ultra-Wide Band (UWB) The range of the UWB system was around 80 m. Due to the limited range, the authors did not recommend UWB systems for WZ open settings. Detection of tagged workers, equipment, and traffic control devices. ehman and Farooq 24 RESEARCH AGENDA
Preceding sections highlight that the existing literature treating CAVs at WZs is limited and still at its early emergence, there are many concepts, functions, and considerations yet to be explored or better understood. This section provides a high-level research agenda for deploying CAV systems at WZ settings. Two main subjects are discussed: establishing broadly a research framework and identifying specifically some major research needs. Figure 9 outlines a proposed framework for researching CAV systems at WZs. Traffic agencies should establish and set major guidelines and standards to be followed by all stakeholders, traffic agencies can still however engage these stakeholders and consider their remarks. The main objective of having generalized standards is to ensure interoperability when advancing CAV systems at WZs. In the absence of well-defined standards, there will be a plenty of CAV systems and functions that cannot be harmonized and integrated, i.e., each equipped-vehicle cannot assure traversing each equipped-WZ compatibly. Three main research themes were defined: traffic operations and performance, infrastructure needs (TCDs and communications), and driver behaviour. The themes should interact with CAV technologies and with each other. Traffic agencies can have the standards and research needs published in a main report that can be supported periodically by revisions or addenda. Table 4 presents the major knowledge gaps and research needs identified based on the present research and available literature. The proposed topics belong to different themes such as mobility, safety, driver’s behaviour, and technology. Table 5 further outlines the available research tools that can be deployed. As mentioned earlier, the literature needs more field-testing and real-world studies. Nevertheless, simulation tools should still play a major role because they can flexibly analyze a multitude of traffic and highway variables and they can also narrow down the scope of the field experiments and their associated cost. Tables 4 and 5 should assist traffic agencies and prospective researchers in identifying and establishing future research efforts.
Figure 9. Framework for researching CAV systems at WZs G ov er n m e n t a l T r a ff i c A g e n c i e s Setting Main Guidelines and Standards Engaging Stakeholders:
CAV Developers TCD Vendors Practitioners and Researchers Contractors
Connected and Automated Vehicles Technologies Traffic Operations and Performance
Mobility Safety Environment
Traffic Control Devices and Communications
Driver Behaviour
CAVs Human-Driven Vehicles
Research Framework Development
Topics that needs further discussion or research Topics that are matured for final revision by traffic agencies Developing Final Standards ehman and Farooq 25
Table 4. Research Needs for Deploying CAV Systems at WZs
No Topic Theme Description
1 Impact of CAV on WZ capacity and travel time
Mobility How to build capacity models that incorporate clustered vehicles, cooperative lane change, dedicated CAV lane, and smart lane flow distribution? What is the impact of MPR? What are the possible merge control strategies that can leverage the shorter-gap, clustering, and synchronized acceleration of CAVs near the WZ taper?
2 Clustering in the WZ vicinity
Mobility When and where clustering should be encouraged, permitted, or prohibited? How clustering strategies can be influenced by each travel lane, WZ configuration, traffic volume, MPR, maximum platoon length, and the distance remaining to the WZ taper? When it is feasible to use a CAV dedicated lane at WZs?
3 Cooperative lane change at WZs
Mobility How to deploy the cooperative lane change concepts at WZs while also considering urgency of the lane change request, WZ configuration, traffic volume, MPR, and communication latency? How much sophisticated the cooperation algorithms and strategies should be? Can we engage human-driven vehicles in this cooperation?
4 Smart lane flow distribution at WZs
Mobility How to define a grouping-based lane change technique whereby drivers change their lanes progressively, i.e., group-after-group, to avoid an all-at-once lane change disruption and in order to achieve an ideal lane flow distribution? What are the possible grouping strategies? How much early upstream of the WZ such system should start drivers’ instruction?
5 CAV environmental impact at WZs
Mobility How much CAV-powered following behaviour, acceleration/deceleration, cooperative lane change, rerouting, and enhanced capacity can reduce emissions at WZs? How environmental benefits differ by the level of connectivity or automation? Would human-driven vehicles be impacted by the surrounding CAVs and change their environmental performance?
6 Heavy vehicles
Mobility and Safety How heavy vehicles will be treated in a mixed traffic stream? What specific functions and concepts must be given to connected and automated trucks at WZs? Where and when truck platooning should be permitted at WZs? How long the platoon should be? Should the lead truck have different automation level as compared to the followers?
7 How to improve existing smart WZ systems with low CAV MPR?
Mobility and Safety Examining how existing smart WZ systems (e.g., queue warning, variable speed limit, alternate route advisory, etc) can benefit from CAV technologies in mix traffic conditions with low MPR.
8 Driver behaviour of human-driven vehicles in a mixed traffic WZ
Driver’s Behaviour Would human-driven vehicles maintain anxiously conservative gap with nearby CAVs or drive more assertively? Would they imitate the behaviour of CAVs without being connected? Would they cooperate?
9 Driver behaviour of CAVs in a mixed traffic WZ
Driver’s Behaviour How drivers will perceive and react to in-vehicle voice, sound, and haptic messages? Can drivers accept tightened following gap between CAVs in a complex and unpredictable WZ geometry or would they alternatively behave conservatively and maintain long gaps? How conservative, normal, and assertive driving behaviours would impact the mobility, safety, and environmental benefits of CAVs?
10 Alignment of vehicle automation levels with safety and mobility benefits at WZs.
Automation Technologies
Mobility and Safety What are vehicle automation levels that can be attainable at WZ areas? How automation level may vary by sub-area (i.e., early upstream, approach area, queuing area, activity area, and termination area)? How to compute automation-induced crash modification factors for each automation level at WZs?
11 Alignment of communication technologies with safety and mobility benefits at WZs.
Communication Technologies
Mobility and Safety What are the best technologies by application? How some factors may influence the selection (e.g., short vs. long term, rural vs. urban, congested vs non-congested WZs, cost, coverage, latency, security, MPR, etc)? Can 5G-cellular technology outperform DSRC? Would a hybrid approach be favourable? What role can smartphones play in the near term?
12 Connecting WZ actors
Communication Technologies What are the best techniques to connect TCDs, workers, signs, and lane markings at WZs? Should one technique be used or should a hybrid approach be adopted? What is the future of robotic at WZs and how they can enhance the work efficiency while maintaining connectivity with vehicles? ehman and Farooq 26
Table 5. Recommended Research Tools for the Future Research Needs
Research Tool or Method Fitting Research Topics (Numbers of Table 4) Opportunities
Microscopic Traffic Simulation Models and Software 1-5, 10, 11
A multitude of traffic and geometric conditions and driver behaviours can be flexibly modelled and examined. Assumptions must be stated clearly in the simulation models in the absence of field calibration.
Driving simulators 3, 5, 8-10
Integrated driving-traffic simulation can be used whereby the driving simulator connects to a microscopic traffic simulation tool which generates the virtual traffic and highway environment. The subject driver and the surrounding traffic continuously exchange their coordinates and speeds and react to each other on real-time basis [e.g., 78-80].
Testbeds (field testing) 5-12
Vehicle-In-The-Loop concept can be used on testbeds whereby the self-driving vehicle moving on the surface connects to a microscopic traffic simulator. The traffic simulator generates virtual traffic and highway environment and communicates actuating attributes to the self-driving vehicle which continuously transmits its trajectory to the simulator on real-time basis [e.g., 81, 82].
Naturalistic Driving Experiments (field testing on real-world highways) 5-12
Naturalistic driving experiments [e.g., 83, 84] can be used on real-world highways whereby the self-driving (or automated) vehicle is equipped with cameras and sensors that measure vehicle performance, driver behaviour and the surrounding conditions. This can be especially useful when a small MPR becomes available on the field. ehman and Farooq 27 CONCLUSIONS
The coexistence of CAVs and WZs is nearing. Integrating CAV technologies and concepts into WZ settings has therefore become an important research direction that is still emerging and raising many interesting knowledge gaps. This paper provided a technical and critical review on this subject aiming to assist in paving the way forward. The paper presented the main concepts, challenges, and opportunities of deploying CAVs at WZs, reviewed the relevant and supporting literature, and outlined a research agenda. The following are the main conclusions of the paper. Because WZ areas involve many traffic challenges, CAVs have the opportunity to credibly showcase their capabilities by responding to these challenges. Among these capabilities are tightened following gap, clustering, synchronized acceleration, early rerouting, lane merge control, cooperative lane change, smart lane flow distribution, dedicated CAV lane, relying on vehicle automation to clear driver’s errors, speed harmonization, and crash warnings. The paper proposed a high-level spatial distribution of these concepts along the WZ area which was subdivided into five segments: further upstream, approach area, queuing area, WZ activity, and termination area. Existing literature only explored limited number of CAV concepts at WZs including tightened following gaps, crash and WZ warnings, and speed harmonization. More efforts are therefore needed and recommended to explore the wide variety of CAV concepts and functions at WZs. The existing literature also relied upon simulation calling for more efforts that can retrieve empirical evidences using test beds for example. Albeit few studies have already assessed some CAV functions at WZs, their results were overall promising and indicated large benefits. These included capacity enhancement, travel time and delay reduction, smoother and safer merges, harmonized speed, enhanced driver’s awareness, and reduced chance of rear-end collision. Nevertheless, some limitations and caveats must be carefully considered when interpreting the results. Mainly, the size of CAV benefits is sensitive to the MPR and the level of traffic congestion. Moreover, the driving behaviour used or assumed in the analysis greatly influences the results; while mobility studies prefer assertive behaviour and short following gaps, safety studies on the other hand tend to assume conservative following behaviour. Finally, excessive use of CAV functions may return adverse impacts or unintended consequences in some circumstances, e.g., excessive CAV rerouting away from WZs may increase the average trip distance and thereby instigate larger exposure to hazards. WZ actors (vehicles, traffic control devices, equipment, and workers) need to be equipped and connected to benefit from CAV functions. Research findings indicate that vehicles can detect signs, pavement marking, and other traffic control devices using computerized vision-based techniques (camera-based and LIDAR); however, for more reliable detection, these elements can be tagged to provide communication-based detection. Many communication techniques and connected traffic devices have been tested positively at WZs. The DSRC is currently the most popular and promising communication technique. However, the rapidly evolving technologies such as cellular communication and constellation of LEO satellites may become other efficient techniques in the near term. The paper proposed a research framework and identified specific research needs for traffic agencies and prospective researchers who are interested in the coexistence of CAVs and WZs. Among the proposed researches are: evaluating the impact of CAVs on WZ capacity and delay, proposing CAV-powered merge control strategies, strategizing CAV clustering and cooperative lane change, developing smart lane flow distribution upstream, assessing CAV environmental benefits at WZs, improving existing smart WZ systems with low CAV MPR, understanding driver’s behaviour in a mixed traffic WZ, analysing the impact of heavy vehicles on connected WZs, aligning the five vehicle automation levels and the several communication technologies with the several mobility and safety benefits, and others. ehman and Farooq 28
CRediT Authorship Contribution Statement
Amjad Dehman:
Conceptualization, Methodology, Investigation, Writing – Original Draft.
Bilal Farooq:
Methodology, Supervision, Writing – Review and Editing, Project Administration.
Acknowledgement
This research was funded and supported by Government of Canada and Lazaret Capital Inc. through the Mitacs Elevate fellowship program.
REFERENCES 1.
Federal Highway Administration (FHWA), “Urban Congestion Trends,” US Department of Transportation, Washington DC, 2012. Cost of Road Congestion in the Greater Toronto and Hamilton Area: Impact and Cost Benefit Analysis of the Metrolinx Draft Regional Transportation Plan. Metrolinx and HDR Corporation, for Greater Toronto Transportation Authority. Final Report, 2008. Schrank, D., B. Eisele, and T. Lomax. Urban Mobility Report 2019, Texas A&M Transportation Institute and INRIX, August, 2019. Federal Highway Administration (FHWA) Work Zone Fact Sheet. 14 th Annual National Work Zone Awareness Week, Federal Highway Administration, 2013. https://ops.fhwa.dot.gov/wz/docs/nwzaw2013factsheet/index.htm. Accessed: August 2020. Nowakowski, C., J. O’Connell, S. Shladover, and D. Cody. Cooperative Adaptive Cruise Control: Driver Acceptance of Following Gap Settings Less Than One Second. Proc., 54th Annual Meeting of the Human Factors and Ergonomics Society, San Francisco, Calif., SAGE Publications, Thousand Oaks, Calif., 2010, pp. 2033–2037. Shladover, S. E., C. Nowakowski, X. Y. Lu, and R. Ferlis. Cooperative Adaptive Cruise Control: Definitions and Operating Concepts. In Transportation Research Record, Journal of the Transportation Research Board, No. 2489, 2015, pp. 145-152. Schakel, W. J., and B. V. Arem. Improving Traffic Flow Efficiency by In-Car Advice on Lane, Speed, and Headway. IEEE Transactions on Intelligent Transportation Systems, Vol. 15, No. 4, 2014. Treiber, M., and D. Helbing. Memory Effects in Microscopic Traffic Models and Wide Scattering in Flow-Density Data. Phys. Rev. E, vol. 68, pp. 046119-1–046119-8, 2003. Zhong, Z., J. Lee, and L. Zhao. Dedicated Lane for Connected and Automated Vehicle: How Much Does A Homogeneous Traffic Flow Contribute?. arXiv:1907.00422v2. 2020, version v2.
Chen, Z., F. He, L. Zhang, and Y. Yin. Optimal Deployment of Autonomous Vehicles Lanes with Endogenous Market Penetration. Transportation Research Part C, No. 72, 2016, pp 143-156.
Rad, S. R., H. Farah, H. Taale, B. V. Arem, and S. P. Hoogendoorn. Design and Operation of Dedicated Lanes for Connected and Automated Vehicles on Motorways: A Conceptual Framework and Research Agenda. Transportation Research Part C, No. 117, 2020.
Ramadan, O., and V. Sisiopiku. Evaluation of Merge Control Strategies at Interstate Work Zones under Peak and Off-Peak Traffic Conditions. Journal of Transportation Technologies, Vol. 06, No. 03, 2016, pp. 118-130.
Tympakianaki, A., A. Spiliopoulou, A. Kouvelas, and M. Papageorgiou. Real-Time Merging Traffic Control for Throughput Maximization at Motorway Work Zones. Procedia - Social and Behavioral Sciences, Vol. 48, 2012, pp. 1545-1556.
LI, T., J. Wu, C. Y. Chan, M. Liu, C. Zhu, W. Lu, and K. Hu. A Cooperative Lane Change Model for Connected and Automated Vehicles. IEEE Access, Volume 8, 03/2020.
Ren, Tianzhu, Y. Xie, and L. Jiang. Cooperative Highway Work Zone Merge Control Based on Reinforcement Learning in a Connected and Automated Environment. In Transportation Research Record, Journal of the Transportation Research Board, Vol. 2674(10), 2020, pp. 363-374. ehman and Farooq 29
An, H., and J. Jung. Design of a Cooperative Lane Change Protocol for a Connected and Automated Vehicle Based on an Estimation of the Communication Delay. Sensors, Volume 18, Issue 10, 2018.
Chambless, J., A.M. Chadiali, J. K. Lindly, and J. McFadden. Multistate work zone crash characteristics. ITE Journal, Institute of Transportation Engineers, 2002, pp. 46–50.
Li, Y., and Y. Bai. Comparison of Characteristics Between Fatal and Injury Accidents in the Highway Construction Zones. Safety Science, No. 46, 2008, pp. 646-660.
Pigman, J.G., and K. R. Agent. Highway crashes in construction and maintenance work zones. In Transportation Research Record 1270, Transportation Research Board, Washington, DC, 1990, pp. 12–21.
Ghiasi, A., X. Li, and J. Ma. A Mixed Traffic Speed Harmonization Model with Connected Autonomous Vehicles. Transportation Research Part C, No. 104, 2019, pp. 210-233.
Learn, S., J. Ma, K. Raboy, F. Zhou, and Y. Guo. Freeway Speed Harmonization Experiment Using Connected and Automated Vehicles. IET Intelligent Transport Systems Journal, Vol. 12, Issue 5, 2018, pp. 319-326.
Garber, N. J., and M. Zhao. Distribution and Characteristics of Crahses at Different Work Zone Locations in Virginia. In Transportation Research Record, Journal of the Transportation Research Board, No. 1794, 2002, pp. 19-25.
Yang, H., K. Ozbay, O. Ozturk, and K. Xie. Work Zone Safety Analysis and Modeling: A State-of-the-Art Review. Traffic Injury Prevention, No. 16, 2015, pp. 387-396.
Sun, H., Z. Zhang, R. Q. Hu and Y. Qian. Wearable Communications in 5G: Challenges and Enabling Technologies. In
IEEE Vehicular Technology Magazine , vol. 13, no. 3, pp. 100-109, 2018.
Choi, B., S. Hwang, and S. Lee. What Drives Construction Workers Acceptance of Wearable Technologies in the Workplace? Indoor Localization and Wearable Health Devices for Occupational Safety and Health. Automation in Construction, No. 84, 2017, pp. 31-41.
Zou, Y. and Qu, X. On the impact of connected automated vehicles in freeway work zones: a cooperative cellular automata model based approach. Journal of Intelligent and Connected Vehicles, Vol. 1 No. 1, 2018, pp. 1-14. https://doi.org/10.1108/JICV-11-2017-0001
Abdulsattar, H., A. Mostafizi, M. R. K. Siam, and H. Wang. Measuring the Impacts of Connected Vehicles on Travel Time Reliability in a Work Zone Environment: An Agent-Based Approach. Journal of Intelligent Transportation Systems: Technology, Planning, and Operations. 2019.
Weng, J., S. Xue, and X. Yan. Modeling Vehicle Merging Behavior in Work Zone Merging Areas During the Merging Implementation Period. IEE Transactions on Intelligent Transportation Systems, Vol. 17, No. 4, April 2016
Ramezani, H., and R. F. Benekohal. Optimized Speed Harmonization with Connected Vehicles for Work Zones. IEEE 18th International Conference on Intelligent Transportation Systems. 2015.
Genders, W., and S. N. Razavi. Impact of Connected Vehicle on Work Zone Network Safety through Dynamic Route Guidance. Journal of Computing in Civil Engineering, ASCE, 2016, 30(2): 04015020.
Abdulsattar, H., A. Mostafizi, and H. Wang. Surrogate Safety Assessment of Work Zone Rear-End Collisions in a Connected Vehicle Environment: Agent-Based Modeling Framework. Journal of Transportation Engineering, Part A: Systems, Volume 144 - Issue 8, 2018.
Qiao, F., R. Rahman, Q. Li, L. Yu, and P. Kuo. Smart Phone Based Forward Collision Warning Message in Work Zones to Enhance Safety and Reduce Emissions. Proceeding of the Transportation Research Board Annual Meeting, 2014, Washington D.C.
Mollenhauer, M., E. White, and N. Roofigari-Esfahan. Design and Evaluation of a Connected Work Zone Hazard Detection and Communication System for Connected and Automated Vehicles (CAVs). Prepared by Safe-D National UTC and Virginia Tech Transportation Institute for the US Department of Transportation. 2019. ehman and Farooq 30
Kamal, U., T. I. Tonmoy, S. Das, and M. K. Hasan. Automatic Traffic Sign Detection and Recognition Using SegU-Net and a Modified Tversky Loss Function with L1-Constraint. IEE Transactions on Intelligent Transportation Systems, Vol. 21, No. 4, 2020.
Yang, Y., H. Luo, H. Xu, and F. Wu. Towards Real-Time Traffic Sign Detection and Classification. IEE Transactions on Intelligent Transportation Systems, Vol. 17, No. 7, 2016.
Tabernik, D., and D. Skocaj. Deep Learning for Large-Scale Traffic-Sign Detection and Recognition. IIE Transactions on Intelligent Transportation Systems, Vol. 21, No. 4, 2020.
Chen, T., and S. Lu. Accurate and Efficient Traffic Sign Detection Using Discriminative AdaBoost and Support Vector Regression. IIE Transactions on Intelligent Transportation Systems, Vol. 65, No. 6, 2016.
Mammeri, A., and A. Boukerche. Inter-Vehicle Communication of Warning Information: An Experimental Study. Wireless Networks, No. 23, 2017, pp. 1837-1848.
Lee, J., Y. W. Seo, W. Zhang, and D. Wettergreen. Kernel-Based Traffic Sign Tracking to Improve Highway Workzone Recognition for Reliable Autonomous Driving. Proceedings of the 16th International IEEE Annual Conference on Intelligent Transportation Systems (ITSC 2013), The Hague, The Netherlands, October 6-9, 2013.
Møgelmose, A., M. M. Trivedi, and T. B. Moeslund. Vision-Based Traffic Sign Detection and Analysis for Intelligent Driver Assistance Systems: Perspectives and Survey. IEEE Transactions on Intelligent Transportation Systems, Vol. 13, No. 4, 2012.
Garcia-Garrido, M. A., M. Ocana, D. F. Florca, M. A. Sotelo, E. Arroyo, and A Llamazares. Robust Traffic Signs Detection by Means of Vision and V2I Communication. 14 th International IEEE Conference on Intelligent Transportation Systems, Washington, DC, USA, 2011.
Gargoum, S., K. El-Basyouny, J. Sabbagh, and K. Froese. Automated Highway Sign Extraction Using Lidar Data. In Transportation Research Record: Journal of the Transportation Research Board, No. 2643, 2017, pp. 1-8.
Rakesh, N., F. Kaiqun, and T. Jason X. Extracting Urban Street Features Using Street Level LiDAR Data for Connected Vehicle Applications. TRB 96 th Annual Meeting Compendium of Papers, Washington DC, 2017.
Tan, M., B. Wang, Z. Wu, J. Wang, and G. Pan. Weakly Supervised Metric Learning for Traffic Sign Recognition in a LIDAR-Equipped Vehicle. IEEE Transactions on Intelligent Transportation Systems, Vol. 17, No. 5, 2016.
Ai, C., and Y. C. J. Tsai. Critical Assessment of an Enhanced Traffic Sign Detection Method Using Mobile LiDAR and INS Technologies. Journal of Transportation Engineering, Volume 141, Issue 5, 2015.
Zhou, L., and Z. Deng. LIDAR and Vision-Based Real-Time Traffic Sign Detection and Recognition Algorithm for Intelligent Vehicle. IEEE 17 th International Conference on Intelligent Transportation Systems (ITSC), Qingdao, China, 2014.
Qiao, F., J. Jia, L. Yu, Q. Li, and D. Zhai. Drivers’ Smart Assistance System Based on Radio Frequency Identification. In Transportation Research Record, Journal of the Transportation Research Board, N. 2458, 2014, pp. 37-46.
Liao, C. F., and M. Donath. Investigating the Effectiveness of Using Bluetooth Low-Energy Technology to Trigger In-Vehicle Messages in Work Zones. By University of Minnesota for Minnesota Department of Transportation Research Services and Library, Report No. MN/RC 2016-38, 2016. http://mndot.gov/research/reports/2016/201638.pdf (Accessed: July 8th, 2020).
Hata, A. Y., and D. Wolf. Feature Detection for Vehicle Localization in Urban Environments Using a Multilayer LIDAR. IEEE Transactions on Intelligent Transportation Systems, Vol. 17, No. 2, 2016.
M. Thuy and F. P. León. Lane detection and tracking based on lidar data. Metrol. Meas. Syst., vol. XVII, no. 3, pp. 311–322, 2010.
S. Kammel and B. Pitzer. Lidar-based lane marker detection and mapping. In Proc. IEEE Intell. Veh. Symp., Eindhoven, The Netherlands, 2008, pp. 1137–1142.
Highway Resource Solutions. Smart Taper. Retrieved on November 2 nd SmartCone Technologies. The Smart Cone Safety Solution. Retrieved on November 2 nd , 2020 from: https://4332cdd4-f736-4a5d-ab15-acba45d49d1f.filesusr.com/ugd/54d203_30d619b86bba44c89a4a0fdd9086cab4.pdf Wang, L., P. Sun, M. Xie, S. Ma, B. Li, Y. Shi, and Q. Su. Advanced Driver-Assistance System (ADAS) for Intelligent Transportation Based on the Recognition of Traffic Cones. Advances in Civil Engineering, Volume 2020, Article ID 8883639, https://doi.org/10.1155/2020/8883639
Shen, X., J. Dumpert, and S. Farritor. Design and Control of Robotic Highway Safety Markers. IEEE/ASME Transactions on Mechatronics, Vol. 10, No. 5, 2005.
Debnath, A. K., R. Blackman, N. Haworth, and Y. Adinegoro. Influence of Remotely Operated Stop-Slow Controls on Driver Behavior in Work Zones. In Transportation Research Record: Journal of the Transportation Research Board, No. 2615, 2017, pp. 19-25.
Trout, N. D., M. D. Finley, and B. R. Ullman. Motorists’ Understanding of Automated Flagger Assistance Devices in Work Zones. In Transportation Research Record: Journal of the Transportation Research Board, No. 2337, 2013, pp. 42-49.
Finley, M. D. Field Evaluation of Automated Flagger Assistance Devices in Work Zones on Two-Lane Roads. In Transportation Research Record: Journal of the Transportation Research Board, No. 2337, 2013, pp. 42-49.
Zeng, X., K. Balke, and P. Songchitruksa. Potential Connected Vehicle Applications to Enhance Mobility, Safety, and Environmental Security. Technical Report No. SWUTC/12/161103-1, Texas Transportation Institute, 2012.
Crash Avoidance Metrics Partnership. Vehicle Safety Communications Project Task 3: Final Report—Identify Intelligent Vehicle Safety Applications Enabled by DSRC. DOTHS-809-859. CAMP Vehicle Safety Communications Consortium, 2005. https://rosap.ntl.bts.gov/view/dot/3925, accessed July 8th, 2020
Maitipe, B. R., U. Ibrahim, M. I. Hayee, and E. Kwon. Vehicle-to-Infrastructure and Vehicle-to-Vehicle Information System in Work Zones: Dedicated Short-Range Communications. In Transportation Research Record, Journal of the Transportation Research Board, No. 2324, 2012, pp. 125-132.
Maitipe, B. R., M. I. Hayee, and E. Kwon. Vehicle-to-Infrastructure Traffic Information System for the Work Zone Based on Dedicated Short-Range Communication: Development and Field Demonstration. In Transportation Research Record, Journal of the Transportation Research Board, No. 2243, 2011, pp. 67-73.
Miao, L., K. Djouani, B. J. Van Wyk, and Y. Hamam. Performance Evaluation of IEEE 802.11p MAC Protocol in VANETs Safety Applications. Proceeding of the IEEE Wireless Communications and Networking Conference (WCNC). 2013, pp. 1681-1686.
Xu, Z., X. Li, X. Zhao, M. H. Zhang, and Z. Wang. DSRC versus 4G-LTE for Connected Vehicle Applications: A Study on Field Experiments of Vehicular Communication Performance. Journal of Advanced Transportation, Volume 17, Article ID 2750452, 10 pages.
Barros, M. T., G. Velez, H. Arregui, E. Loyo, K. Sharma, A. Mujika, and B. Jennings. CogITS: Cognition-Enabled Network Management for 5G V2X Communication. IET Intelligent Transport Systems, Volume 14, Issue 3, 2020.
Ansari, S., J. Ahmad, S. A. Shah, A. K. Bashir, T. Boutaleb, and S. Sinanovic. Chaos-based Privacy Preserving Vehicle Safety Protocol for 5G Connected Autonmous Vehicle Networks. Transactions on Emerging Telecommunications Technologies. Volume 31, Issue 5, 2020.
Do, D-T., T-TT. Nguyen, C-B. Le, and J. W. Lee. Two-Way Transmission for Low-Latency and High-Reliability 5G Cellular V2X Communications. Sensors, Vol 20(2), 2020. ehman and Farooq 32
P. Wang, B. Di, H. Zhang, K. Bian and L. Song. Platoon Cooperation in Cellular V2X Networks for 5G and Beyond.
IEEE Transactions on Wireless Communications , vol. 18, no. 8, pp. 3919-3932, 2019, doi: 10.1109/TWC.2019.2919602.
Ge, X., H. Cheng, G. Mao, Y. Yang, and S. Tu. Vehicular Communications for 5G Cooperative Small-Cell Networks. IEEE Transactions on Vehicular Technology, Vol. 65, No. 10, 2016.
Reid, T. G. R., A. M. Neish, T. Walter, and P. K. Enge. Broadband LEO Constellation for Navigation. Journal of the Institute of Navigation, Vol. 65, No. 2, 2018.
Ge, H., B. Li, L. Nie, M. Ge, and H. Schuh. LEO constellation optimization for LEO enhanced global navigation satellite system (LeGNSS). Advances in Space Research, Volume 66, Issue 3, 2020.
Iannucci, P. A., and T. E. Humphreys. Fused Low-Earth-Orbit GNSS. arXiv:2009.12334. 2020, version v1.
Reid T. G. R., B. Chan, A. Goel, K. Gunning, B. Manning, J. Martin, A. Neish, A. Perkins, and P. Tarantino. Satellite Navigation for the Age of Autonomy. Proceedings of the 2020 IEEE/ION Position, Location and Navigation Symposium (PLANS), Portland, OR, USA, 2020, pp. 342-352, doi: 10.1109/PLANS46316.2020.9109938.
Zaman, A. U., M. I. Hayee, N. Katta, and S. Mooney. Traffic Information System to Deliver In-Vehicle Messages on Predefined Routes: Use of Dedicated, Short-Range Vehicle-to-Vehicle Communication. In Transportation Research Record: Journal of the Transportation Research Board, No. 2559, 2016, pp. 73-80.
Ibrahim, U., M. I. Hayee, and E. Kwon. Hybrid Work Zone Information System with Portable Changeable Message Signs and Dedicated Short-Range Communication. In Transportation Research Record: Journal of the Transportation Research Board, No. 2380, 2013, pp. 29-35.
Azadi, F., Y. Adu-Gyamfi, C. Sun, and P. Edara. Mobile Applications Development and Testing for Work Zone Activity Real-Time Data Collection. In Transportation Research Record, Journal of the Transportation Research Board, Vol. 2674(6), 2020, pp. 351-362.
Zhao, Y., A. Wagh, Y. Hou, K. Hulme, C. Qiao, and A. W. Sadek. Integrated Traffic-Driving-Networking Simulator for the Design of Connected Vehicle Applications: Eco-Signal Case Study. Journal of Intelligent Transportation Systems, No. 20(1), 2016, pp. 75-87.
Sun, J., Z. Ma, T. Li, and D. Niu. Development and Application of an Integrated Traffic Simulation and Multi-Driving Simulators. Simulation Modelling Practice and Theory, No. 59, 2015, pp. 1-17.
Hou, Y. Y. Zhao, K. F. Hulme, S. Huang, Y. Yang, A. W. Sadek, and C. Qiao. An Integrated Traffic-Driving Simulation Framework: Design, Implementation, and Validation. Transportation Research Part C, No. 45, 2014, pp. 138-153
Fayazi, A., A. Vahidi, and A. Luckow. A Vehicle-in-the-Loop (VIL) verification of an all-autonomous intersection control scheme. Transportation Research Part C, No. 107, 2019, pp. 193-210.
Horváth, M. T., Q. Lu, T. Tettamanti, A. Török, and Z. Szalay. Vehicle-In-The-Loop (VIL) and Scenario-In-The-Loop (SCIL) Automotive Simulation Concepts from the Perspectives of Traffic Simulation and Traffic Control. Transport and Telecommunication, Vol. 20, No. 2, 2019, pp. 153-161.
Bharadwaj, N., P. Edara, and C. Sun. Risk Factors in Work Zone Safety Events: A Naturalistic Driving Study Analysis. In Transportation Research Record, Journal of the Transportation Research Board, Vol. 2673(1), 2019, pp. 379-387.84.