Effects of connected and autonomous vehicle merging behavior on mainline human-driven vehicle

Lishengsa Yue; Mohamed Abdel-Aty; Zijin Wang

doi:10.1108/jicv-08-2021-0013

Effects of connected and autonomous vehicle merging behavior on mainline human-driven vehicle

Yue, Lishengsa; Abdel-Aty, Mohamed; Wang, Zijin 2022-02-17 00:00:00 Purpose – This study aims to evaluate the inﬂuence of connected and autonomous vehicle (CAV) merging algorithms on the driver behavior of human-driven vehicles on the mainline. Design/methodology/approach – Previous studies designed their merging algorithms mostly based on either the simulation or the restricted ﬁeld testing, which lacks consideration of realistic driving behaviors in the merging scenario. This study developed a multi-driver simulator system to embed realistic driving behavior in the validation of merging algorithms. Findings – Four types of CAV merging algorithms were evaluated regarding their inﬂuences on driving safety and driving comfort of the mainline vehicle platoon. The results revealed signiﬁcant variation of the algorithm inﬂuences. Speciﬁcally, the results show that the reference-trajectory-based merging algorithm may outperform the social-psychology-based merging algorithm which only considers the ramp vehicles. Originality/value – To the best of the authors’ knowledge, this is the ﬁrst time to evaluate a CAV control algorithm considering realistic driver interactions rather than by the simulation. To achieve the research purpose, a novel multi-driver driving simulator was developed, which enables multi-drivers to simultaneously interact with each other during a virtual driving test. The results are expected to have practical implications for further improvement of the CAV merging algorithm. Keywords Driving simulator, Connected and autonomous vehicle, Merging algorithm, Merging behavior, Safety and comfort, Driving safety, Driving comfort Paper type Research paper merging path generation algorithm and social-psychology- 1. Introduction based merging algorithm. For the ﬁrst type, the objective of Connected and autonomous vehicle (CAV) technology has this type of merging control is to ﬁnd a reference trajectory been gaining more and more attention in recent years; it to guide the autonomous vehicle by considering physical releases drivers from heavy driving tasks and avoids driver and kinematic restrictions to fulﬁll a successful merging. Lu errors. One challenge of CAV technology is its adaptability and Hedrick (2003) and Lu et al.(2004) designed a set of in critical trafﬁc scenarios. A typical critical scenario is the kinematical restriction functions in terms of the acceleration merging scenario at the freeway ramp area; it is the hotspot and position, to ensure that the vehicles can reach the of trafﬁc crashes. In total, 18% of all interstate freeway merging pointsatan appropriate time. Wang et al. (2013) crashes, 17% of the injury crashes and 11% of the fatal proposed a cooperative driving algorithm based on vehicular crashes occurred at interchanges, and most proportion of operation characteristics for the ramp merging. They these crashes took place at the entrance or exit ramps considered the position and speed requirements for both (Ahammed et al., 2008; McCartt et al.,2004). Given that one and two vehicles on the ramp. The second type there is usually signiﬁcant vehicle interaction at the merging considers driving preference regarding the merging behavior areas, the design of the CAV merging algorithm is critically important; the algorithm is supposed to ensure a safe © Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang. Published in merging behavior; meanwhile, it is expected to disturb the Journal of Intelligent and Connected Vehicles. Published by Emerald mainline driving as little as possible. Publishing Limited. This article is published under the Creative Commons Many studies have designed CAV merging control Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, algorithms. Generally, the merging algorithms can be translate and create derivative works of this article (for both commercial categorized into two types: physical-restriction-based and non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence maybe seen at http:// creativecommons.org/licences/by/4.0/legalcode The current issue and full text archive of this journal is available on Emerald The authors acknowledge the ﬁnancial support of the Innovative Insight at: https://www.emerald.com/insight/2399-9802.htm Technology Administration of US Department of Transportation, Award No. DTRT13-G-UTC53 (SAFER-SIM). Received 19 August 2021 Journal of Intelligent and Connected Vehicles Revised 18 October 2021 5/1 (2022) 36–45 8 November 2021 Emerald Publishing Limited [ISSN 2399-9802] [DOI 10.1108/JICV-08-2021-0013] Accepted 23 November 2021 36 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 and manipulates the merging model based on factors such as CarMaker, TFSS and Matlab/SIMULINK is often used desired gap distance and desired mainstream speed. (Madhusudhanan, 2019; Nalic et al., 2020a; Nalic et al., Basically, this type of algorithm deﬁnes the model in a form 2020b). of preferred and actual accelerations and distance gap However, the lack of realistic driver behavior in the algorithm maintain (Awal et al., 2013a, 2013b; Chou et al., 2016; validation deteriorates the credibility of the algorithm. As Karimi et al., 2020; Wu et al., 2019). An advanced form of pointed by Aksjonov et al. (2020), purely computation thesecond typeis to consider multipleoptimization targets simulation does not guarantee realistic environments for a and generate the reference merging path by solving the testing vehicle, and this is the reason that in recent years the optimal solution. Ding et al. (2019) proposed a rule-based concept of “hardware-in-the-loop” or “human-in-the-loop” merging algorithm with minimizing travel time and delay as becomes more prevalent (Aksjonov et al., 2020; Artunedo et al., the target; they formed a closed-form analytical solution to 2015; Dang et al., 2020; Schreiber et al.,2018). Basically, using real hardware or a driver to test the algorithm is more reliable achieve a near-optimal merging sequence. Letter and Elefteriadou (2017) presented a longitudinal freeway than using a driver behavior model particularly when it is merging control algorithm, which used the average travel necessary to observe vehicle interactions and possible improper speed as the optimization target, and they used LINGO to driving adaptation behaviors such as driving errors and resolve the optimal solution. aggressive driving. Theoretically speaking, a driver behavior The veriﬁcation of the above algorithms is usually based on model is controlled by many kinematical restrictions with the simulation platforms, either through a single platform or a co- purpose of generating “smooth” behaviors, and it is hard to simulation based integrated simulation platform. Some single mimic improper driving adaptations (Nalic et al., 2020a). platforms are basically microscopic trafﬁc ﬂow simulation Regarding the CAV merging control algorithms, most of the software (TFSS) such as Vissim and SUMO; they have aforementioned studies were based on simulation, which is embedded driving behavior models, which can mimic car- hard to represent realistic driver behaviors in the merging scenario. Very few of them conducted ﬁeld testing; however, following or lane change behaviors. For example, Vissim used a rule-based algorithm to initiate lateral lane change behavior and because of the safety consideration, only conservative a psychophysical model for the longitudinal car-following algorithms and restricted testing conditions (such as low movement (Fellendorf and Vortisch, 2001). Different from driving speed) were tested. Many studies proved that driver Vissim, SUMO used a Krauss model as its default car- behavior can signiﬁcantly affect the crash and safety level at the following model (Bieker-Walz et al.,2017). However, merging area. Weng et al. (2015) found that the drivers’ individual vehicle dynamics modeling is not a strength for these merging behavior is highly correlated with the rear-end crash platforms. On the contrary, some other single platforms, such risk; there will be high rear-end crash risks when the merging as CarMaker, have a better ability of modeling vehicle dynamic vehicle travels at either a very high or low speed. Weng and details such as powertrain and sensor system. The driver Meng (2014) found that if the merging action initiates earlier, behavior model used in the CarMaker is based on a there will be a lower rear-end crash potential. Reinolsmann proportional-integral-derivative controller, with considerations et al. (2019) also suggested earlier lane change because it can of psychological studies and measurements from real test contribute to smooth maneuvers and gradual speed reductions drivers (Olofsson and Pettersson, 2015). Whatever driver particularly at the rural expressway ramp area. Moreover, Potzy behavior models they are, the models result from speed, speed et al. (2019) concluded that drivers on the mainline prefer an difference, distance gap, vehicle dynamic restrictions or efﬁcient lane change of the autonomous vehicle on the ramp, individual driver characteristics. and results show that drivers would tolerate less compliance Integrated simulation platforms provide more explicit with safety distance to have less interacting trafﬁc. It is quite simulation regarding individual vehicle dynamics or a better necessary to consider the realistic driver behavior for CAV power to optimize CAV control algorithms during the merging control algorithms so that the CAV merging behavior running time; the integrated platform can consist of a TFSS can be more acceptable and predictable for mainline drivers. and several other simulators such as IPG CarMaker and Therefore, the objective of this study is to evaluate the Matlab/SIMULINK. Basically, the TFSS is more efﬁcient inﬂuence of CAV merging algorithms on driver behavior of at simulating microscopic trafﬁc ﬂow while it ignores the human-driven vehicles on the mainline, by using the human-in- vehicle dynamic details; on the contrary, the IPG CarMaker the-loop concept. Several classical merging algorithms were can better simulate the vehicle dynamics such as the tested in this study, representing the physical-restriction-based powertrain system and sensor system; while the Matlab/ merging path generation algorithm and social-psychology- SIMULINK is mainly for algorithm optimization purpose, based merging algorithm; then their inﬂuence on the mainline its embedded mathematical toolboxes can be used to resolve trafﬁc was analyzed given their algorithm framework features. optimization problems and generate optimized parameters/ This study is expected to conclude design principles of merging outputs of control algorithms. When complicated trafﬁc algorithms that have less inﬂuence on the mainline trafﬁc. The ﬂow simulation is not necessary, IPG CarMaker is often driving safety and driving comfort of mainline human drivers used to verify CAV control algorithms, such as longitudinal would be analyzed to distinguish the performance of different cruise control (Kuutti et al., 2019), lateral lane change merging algorithms. To account for crash risks in a realistic (Samiee et al., 2016), overtaking path planning (Nguyen ﬁeld testing, a driving simulator experiment would be used et al., 2017) and tactical behavior planning (Sefati et al., instead. 2017); while the CAV algorithm is required to be tested in This study is organized as follows: CAV Merging Algorithm certain trafﬁc ﬂow conditions, a co-simulation between section 2 introduces classical CAV merging algorithms that 37 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 were tested in this study; Experimental Design section 3 based on the optimization model was not investigated in this presents the multi-driver simulator system developed for this study, because of its computation cost in the real-time driving study, experimental scenarios and data analysis method; simulator system. These tested merging algorithms and their Results section 4 presents the results and Discussion section 5 examples are listed below. investigates the results; ﬁnally, the Limitation section 6 Reference-trajectory-based merging algorithm. This study presents the study limitation and Conclusion section 7 adopted the algorithm (denoted as AHS) and its parameters proposed by Lu and Hedrick (2003) and Lu et al. (2004) as an summarizes this study. example, which deﬁned the reference trajectory v (t)as md follows: 2. Connected and autonomous vehicle merging algorithm v t ¼ vt ; > mdðÞ merg ðÞ merg Almost all CAV merging algorithm designs adopt a concept of < 1 a t vt 1 a t vðÞ t 1 ðÞ ðÞ i ðÞ merg ðÞ i p i v ðÞ t ¼ when t ðÞ t 1 iDt T ; (1) “virtual platoon,” with a connotation of projecting the CAV’s md i merg merg virt vðÞ t 1 position from the merging ramp to the mainline, and generating p i when T <ðÞ t 1 iDt T ; virt merg merg a “virtual platoon” consists of both human-driven vehicles (yellow) and projected CAV (gray) (Figure 1). The projection is exactly based on geometric parameters of the CAV, and it a t ¼ a t ; b > 0 (2) ðÞ ðÞ i i determines the relative position of the projection vehicle to other mainline vehicles. The core idea of the CAV merging algorithm is to manipulate the speed and acceleration of the v t 1 Dt ðÞ p j projected CAV, so that it can maintain a safe headway distance j¼1 a t ¼ (3) 0ðÞ i X to the front vehicle under a desired traveling speed. Basically, in vðÞ t 1 Dt1 dist para p j a fully connected and automated environment, a central j¼1 controller will be set up covering the upstream and downstream of merging area, and collect speed and location information of where v(t) is the merging vehicle speed, v (t) is the speed of the all vehicles (both on mainline and ramp) entering the control ﬁrst vehicle in the platoon on mainline, t is the time when merg area; the vehicles in the control area will be manipulated so that the merging algorithm starts, T is the time when the virtual virt the speed and headway distance of each single vehicle in the platoon is established but merging is not complete yet, T is merg virtual platoon can be well accommodated. Speciﬁcally, the time when the merging is ﬁnished, b is a coefﬁcient and the controller will accommodate the autonomous vehicle dist_para is the initial distance relationship between vehicles (projected) based on its relative speed and position to the considering desired distance before and after the merging. leading vehicle (ﬁrst vehicle in the platoon), and consecutively Detailed variable deﬁnitions can be found in Lu and Hedrick accommodate the second and third vehicles based on similar (2003) and Lu et al.(2004). safety considerations toward the vehicles in front of them. Social-psychology-based merging algorithm. This study tested In this study, the driving environment is partially connected, three examples, which were borrowed from car-following and only the CAV can manage its movement by sensing the models, by considering the driver’s desire to main a certain leading vehicle; for the second and third vehicles, drivers need to speed and distance to the leading vehicle. The intelligent driver determine the driving by themselves rather than follow the central model (IDM) (Treiber et al.,2000), the generalized force controller. Therefore, the controller, which is embedded with the model (GFM) (Helbing and Tilch, 1998) and the k-leader fuel- merging algorithm, will take the information of the ﬁrst vehicle as efﬁcient (KLFE) model (Awal et al., 2013a, 2013b) were used input to manage the movement of the CAV. as examples in this study. The model parameters were identical Two types of classical CAV merging algorithms, the to the ones in corresponding studies. reference-trajectory-based merging algorithm generation The IDM is given by: algorithm and the social-psychology-based merging algorithm, "# were reproduced in this study based on previous studies. The v S ðÞ v;Dv desire a ¼ A 1 (4) study veriﬁed their effects on the vehicle platoon on the V S mainline. The third type of CAV merging algorithm that is rﬃﬃﬃﬃﬃ v vDv Figure 1 Virtual platoon and projected CAV S ðÞ v; Dv ¼ g 1 g 1 vT 1 pﬃﬃﬃﬃﬃﬃ (5) desire 0 1 2 Ab where a is the suggested acceleration; v and V are the current speed and desired speed, respectively; Dv is the speed difference to the preceding vehicle; S and S are the current desire following distance and desired following distance, respectively; A and b are maximum desired acceleration and deceleration, respectively; g and g are different jam distance parameters 0 1 and d is a constant coefﬁcient. Detailed variable deﬁnitions can be found in Treiber et al.(2000). The GFM is given by: 38 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 sns ðÞ v desire the preceding vehicle; T is the safe time headway; S is the jam Ra V 1 e v a ¼ b (6) distance. Detailed variable deﬁnitions can be found in Awal et al.(2013a, 2013b). It is worth mentioning that the KLFE considers vehicles on both the ramp and mainline. s s ðÞ v desire ðÞ d u Dv Dve b ¼ (7) 3. Experimental design 3.1 Apparatus A multi-driver driving simulator system (Figure 2)wasdeveloped S ðÞ v ¼ s 1 vT (8) desire 0 to test a vehicle platoon in a virtual driving scenario. Compared with a realistic ﬁeld test, the advantage of using the driving where a is the nth vehicle’s acceleration; v and V are the current simulator isthatitcan test dangerousdriving scenarioswithout speed and desired speed, respectively; Dv is the speed difference real collision risks. The simulator system designed a data to the preceding vehicle; s , s and s are the current following n desire 0 collection module, a vehicle physics module, a scenario distance, desired safe following distance and minimum following management module and a communication module. The data distance, respectively; u is the Heaviside function; T is the safe collection module collects driver behavior data in the scenario, time headway; t and t are the acceleration time and braking a d such as brake, throttle, steering wheel, speed and position; the time, respectively; R and R arethe rangeofthe acceleration and a d vehicle physics module simulates vehicle dynamics and related range of the braking interaction, respectively. Detailed variable physical features, such as engine dynamics and collision effects; deﬁnitions can be found in Helbing and Tilch(1998). the scenario management module conﬁgures scenario control The KLFE is given by: scripts, and it manages all types of scenario objects and their ðÞ ðÞ v t1Dt ¼ min v t1Dt (9) actions; as for the communication module, it connects multiple n n;m driver clients and distributes the simulation data between clients simultaneously (Figure 3). a safe ðÞ v t1 Dt ¼ max 0; min v ðÞ t ; v ðÞ t (10) n;m n;m n;m 3.2 Merging scenario design This study designed a merging scenario as illustrated in v ðÞ t ¼ v ðÞ t 1 k (11) n;m Figure 4. Three human-driven vehicles (the ﬁrst to third yellow vehicles) are traveling on the mainline, and they form a stable 2 3 4 vehicle platoon. A CAV (fourth red vehicle) is merging into the desire v ðÞ t s ðÞ t v ðÞ t n n n;m 4 5 mainline from a ramp, and it is supposed to cut in between the k ¼ ADt 1 (12) V V n;m ﬁrst and second vehicles in the platoon; the CAV is controlled by the automatic merging algorithm. The second human- driven vehicle determines whether to yield to the CAV, based hi 2 0 v ðÞ t p ¼ 2 x ðÞ t x ðÞ t l v ðÞ t Dt (13) on the safety consideration. Normally, the CAV would appear m n n ahead in the second human-driven vehicle’s view; therefore, the second human-driven vehicle would slow down. However, in bDt1qif q 0 safe rare cases, the second human-driven vehicle decides to v ðÞ t ¼ (14) n;m bDt1 v ðÞ t if q < 0 accelerate and overtake the CAV; in these cases, the merging algorithm will recognize that the second human-driven vehicle qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ would arrive at the merging point before the CAV, thus the ðÞ q ¼ b Dt bp (15) algorithm will change its goal to following the second human- driven vehicle. The study arranged a ﬁfth environmental vehicle (blue) in front of the vehicle platoon in the mainline; the desire s ðÞ t ¼ S 1 b ðÞ n m (16) n;m vehicle follows a predeﬁned path and speed, and it is used as a v ðÞ t v ðÞ t v ðÞ t n n m Figure 2 Multi-driver driving simulator system framework b ¼ max 0; v ðÞ t T 1 pﬃﬃﬃﬃﬃﬃﬃﬃﬃ (17) 2 Ajbj where n and m stand for the nth and mth vehicles in the platoon; desire s is the desired following distance between the nth and mth n;m vehicles; v is the following vehicle speed when the distance n;m between the following and preceding vehicle is large, whereas safe the v is the following vehicle safe speed when the gap distance n;m is small; v , v and V are the current following speed, current n m preceding speed and desired speed, respectively; x and x are n m the following and preceding vehicle positions; l is the effective size plus a margin; A is the maximum desired acceleration; b is the maximum braking rate; b is the estimated braking rate of 39 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 group of drivers was given a practice driving so that they can be Figure 3 Multi-driver driving simulator system and experiment familiar with the driving environment. During the formal experiment, each group drove four tracks, and each track had one type of merging algorithm repeated six times. In other words, the drivers experienced ﬁve merging scenarios in one track. 3.4 Inﬂuence period during merging It is worth mentioning that during the above “merging scenario,” the CAV and second vehicle would not interact with each other all the time. Because of the sight of the second driver’sview, only when the CAV is close enough, the second driver would be affected and accommodate driving behaviors to the merging vehicle. This period is deﬁned as the inﬂuence period. Given that the driving behaviors are assumed to be different between the normal and inﬂuenced driving periods, the ﬁnite Gaussian mixture model (GMM) was used to distinguish two periods in the merging scenario. The GMM assumes that data of different features is coming from a mixture of two or more Gaussian distributions (i.e. clusters), and the GMM allocates data points to most probable distributions by expectation- maximization (EM) algorithm (Scrucca et al., 2016). Similar trajectory clustering practice was conducted by Mohammed et al.(2019);the researchers used ﬁnite GMM to cluster cyclists overtaking and following trajectories into different states. In the merging scenario, the second driver manages the throttle and brake to maintain the safety buffer between both the ﬁrst vehicle and CAV. Therefore, this study used second vehicle’s speed, throttle and brake positions and distances to reference vehicle to control the speed of the vehicle platoon, in the ﬁrst vehicle and CAV as trajectory features to be clustered. case either too fast or too slow. Figure 4 shows an example of the throttle clusters generated by In each experiment, a set of three connected drivers drove GMM. It shows that at the time of around 100, the second through a track that contains ﬁve merging scenarios. These ﬁve driver notices the merging CAV and begins to monitor the scenarios have the same merging algorithms, and the driver’s collision risk; then at the time of around 160, the second driver average driving performance was analyzed later. The begins to release the pedal position to slow down; during the experiment was a within-subjects experiment, and three drivers time point of 100–160, the throttle position does not change, experienced all ﬁve merging algorithms; therefore, in total, ﬁve this might be because of the driver’s reaction time delay. It is experiments were conducted. The merging algorithms were worth mentioning that the driving periods of the second driver presented in a randomized way to account for the order effect were applied to the third driver in this study (Figure 5). (Yue et al., 2020). In the experiment, the distance to the front vehicle and its speed information was displayed on the following vehicle’s screen; referring to this information display, 3.5 Driving performance metrics each driver was asked to follow the front vehicle keeping a This study mainly investigated the inﬂuence of CAV merging distance of 60–80 m; the speed limit was 55 mph. algorithms on the second and third human-driven vehicles. Two driving periods were analyzed: the merging and following 3.3 Experimental procedure periods. The merging period is deﬁned as the period from the In total, 16 groups of drivers (i.e. 48 drivers) conducted the experiment. They were driving on the mainline and an Figure 5 Throttle position during merging scenario autonomous car controlled by the merging algorithm was merging from the ramp. Before the formal experiment, each Figure 4 Merging scenario in experiment 40 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 time the CAV begins to move on the ramp to the time it arrives Nandi et al.,2015). In the following period, the minimum at the merging point, which occurs in the merging scenario. headway distance was used to indicate the driving comfort level The following period (Figure 6) is from the time that (Bellem et al.,2016); the larger the headway distance, the more the CAV ﬁnishes merging and begins to drive with the vehicle comfortable the driving is. platoon, to the time that the CAV leaves the lane of vehicle platoon (the leaving point was given); the following actions 4. Results speciﬁcally refer to the actions of second human-driven vehicle 4.1 Inﬂuence on driving safety (following the CAV) and third human-driven vehicle (following During the merging period, for the second driver, both the the second human-driven vehicle). minimum TTC and DRAC between the four merging For each driving period, two aspects of metrics were algorithms were signiﬁcantly different (F = 5.62, p-value < collected in terms of both driving safety and driving comfort. 0.01 and F = 14.45, p-value < 0.01, respectively). Figure 7 The driving safety measures include the minimum time-to- shows that the AHS algorithm had the largest minimum TTC, collision (TTC) and the deceleration rate to avoid crashes which was 19.45 s; signiﬁcantly smaller than the AHS (DRAC). The TTC was introduced by Hayward(1972) and is algorithm, the GFM algorithm had a minimum TTC of only the most widely used surrogate safety measure (SSM); it 6.49 s. As for the IDM and KLFE algorithms, their minimum indicates the time that is left for the following vehicle to hit a TTC values were between AHS and GFM; the KLFE had a leading vehicle. The TTC is given by: slightly higher safety level than the IDM. In addition, the AHS algorithm had a very small DRAC of only 0.007; the KLFE’s S L ; v > v 2 1 DRAC was slightly larger, i.e. 0.009, whereas the GFM and TTC ¼ (18) v v 2 1 IDM had a much larger DRAC of around 0.7. For the third 1; otherwise driver, only the minimum TTC was found signiﬁcantly different between merging algorithms (F = 2.45, p-value = where S is the distance gap between the leading and following 0.056). Similar to the inﬂuence on the second driver, the AHS vehicles; L is the vehicle length; v and v are speeds of 2 1 had the largest minimum TTC, which was 36.77 s; the GFM following and leading vehicles, respectively. TTC can be had the smallest minimum TTC of only 22.16 s. The level of calculable only when v is greater than v . In this study, TTC 2 1 minimum TTC of the IDM and KLFE was between AHS and was assigned a large value of 100 when v < v . Obviously, a 2 1 GFM. large value of minimum TTC indicates a high safety level. During the following period, the second driver’s minimum The DRAC is also a widely used SSM, proposed by Cooper TTC was found signiﬁcantly different between four types of and Ferguson (1976); it indicates the required minimum merging algorithms (F = 2.31, p-value = 0.068). The AHS had deceleration rate for a following vehicle to avoid a crash with a the largest minimum TTC of 41.30 s, whereas the KLFE had leading vehicle. The DRAC is given by: the smallest minimum TTC of 30.23 s. The GFM was slightly 2 larger than KLFE, which was 36.20 s, and the IDM had a v v ðÞ 2 1 ; v > v minimum TTC of 38.40 s. Similar to the second driver, the 2 1 DRAC ¼ (19) S L third driver had signiﬁcantly different minimum TTC between 0; otherwise merging algorithms (F = 3.81, p-value < 0.01). Speciﬁcally, the AHS had the largest minimum TTC of 11.61 s, whereas the Usually, a threshold DRAC would be selected and the GFM and IDM had the smallest minimum TTC of around percentage of DRAC > DRAC is used to indicate the safety 8.35 s. The minimum TTC of KLFE was 9.58 s. level; a large percentage represents a dangerous situation. In this study, a threshold DRAC of 3.0 s was adopted, and the 4.2 Inﬂuence on driving comfort “DRAC” term mentioned in later sections is the percentage During the merging period, the second driver’s mean value, weighted by trajectory length. Both the minimum TTC deceleration was signiﬁcantly different between the four and DRAC were applied to the merging and following periods. merging algorithms (F = 4.74, p-value = 0.002). Figure 8 In terms of the driving comfort measures, the mean shows that the AHS had the smallest deceleration of 2.64 m/s , deceleration/acceleration and the average jerk during the whereas the GFM had the largest deceleration of 4.23 m/s . deceleration/acceleration processes were used for the merging The decelerations of the KLFE and IDM were in between with period. The jerk is the derivative of deceleration/acceleration. In a value of 3.38 and 3.11 m/s , respectively. The average jerks the merging period, these measures were proved to have a negative during the deceleration process were also found signiﬁcantly relationship with driving comfort (Bellem et al., 2016, 2018; different between the merging algorithms (F = 14.45, p-value < 0.01). To be speciﬁc, the GFM had the largest average jerk Figure 6 Following period when a CAV cuts in between the ﬁrst and during the deceleration process, which was 36.86; the IDM and second vehicles AHS had a relatively smaller average jerk value close to each other of around 27.50; the jerk value of the KLFE was in between and it was 30.04. In terms of the minimum headway distance, a signiﬁcant difference was found because of the merging algorithms (F = 62.26, p-value < 0.01). The AHS and KLFE had a very large minimum headway distance of 37.99 and 36.11 m, respectively, whereas the GFM had the smallest 41 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 minimum headway distance of 17.57 m. The IDM had a the other three algorithms are very close. In terms of the minimum headway distance of 23.64 m, slightly larger than the minimum headway distance, the AHS had the largest one of GFM. For the third driver, the average jerks during the 53.59 m, whereas the other three algorithms are very close. acceleration process were found signiﬁcantly different between the merging algorithms (F = 2.21, p-value = 0.08). The AHS 5. Discussions had the largest average jerk of 47.80, whereas the KLFE had The heterogeneity of CAV merging behavior caused the smallest average jerk of 36.83. The average jerks of the signiﬁcantly different inﬂuences on the mainline vehicle GFM and IDM were close to each with a value of 39.62 and platoon. For the second driver, regarding the driving safety in 40.48, respectively. either merging or following periods, the minimum TTC and During the following period, the second driver, both the DRAS show that AHS had the least negative inﬂuence, whereas mean acceleration and deceleration were found to be the GFM had the most negative inﬂuence. Regarding the signiﬁcantly affected by the merging algorithm. For mean driving comfort, the deceleration, jerk and headway distance acceleration, the GFM had the largest one of 4.33 m/s , show that the AHS had the best driving comfort, whereas the whereas the AHS had the smallest one of 3.34 m/s .The GFM had the worst driving comfort in both two driving mean accelerations of IDM and KLFE were 3.79 and periods. The inﬂuence of IDM and KLFE on driving safety and 3.77 m/s , respectively. Additionally, the GFM had the comfort was generally in between the AHS and GFM. largest mean deceleration of 3.47 m/s , whereas AHS had Compared with the second driver, for the third driver, the the smallest mean deceleration of 2.89 m/s .The mean heterogeneity of inﬂuence of CAV merging behaviors deceleration of the IDM and KLFE was close to each other. was much less signiﬁcant. In terms of driving safety, the DARC In terms of the average jerk, it was found to be signiﬁcantly was not signiﬁcantly different and only the minimum TTC was affected by the merging algorithms during the acceleration observed to be varied between CAV algorithms. The minimum process (F = 4.92, p-value < 0.01). The GFM had the TTC shows that in both merging and following periods, the largest one of 46.81, whereas the AHS had the smallest one AHS was the safest, whereas the GFM was the riskiest one. In of 41.93. The average jerks of IDM and KLFE were 43.29 terms of driving comfort, in the merging period, the average and 43.72 m/s , respectively. For the minimum headway jerk shows that the AHS was the least comfortable one, and distance, a signiﬁcant difference was also found (F = 47.21, both the GFM and IDM had a better comfortable level than the p-value < 0.01). The GFM had the smallest one of 21.60 m, whereas the AHS had the largest one of 41.01 m. The KLFE AHS; the KLFE was the most comfortable one. In the also had a large minimum headway distance of 40.69 m, following period, the mean acceleration shows that the GFM was the least comfortable one, whereas the minimum headway whereas the value of IDM was 32.07 m. distance shows that the AHS is the most comfortable one. During the following period, the third driver’s mean The results show that the AHS can guarantee driving acceleration and minimum headway distance were found to be safety and driving comfort among the tested four types of signiﬁcantly affected by the merging algorithm (F = 2.43, p- value = 0.06 and F = 4.11, p-value < 0.01, respectively). The merging algorithms; nevertheless, its driving comfort may GFM had the largest mean acceleration of 4.51 m/s , whereas deteriorate for the later part of the vehicle platoon in the Figure 7 Driving safety level of automatic merging algorithm 42 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 merging period. It was interesting that the AHS, as a GFM was the worst merging algorithm because it caused reference-trajectory-based merging algorithm, can achieve more negative effects on driving safety and driving a comprehensive good performance in terms of both comfort; this might be because it is not much suitable to driving safety and comfort, without considering many capture a driver’s comprehensive driving preference. other optimization goals. In addition, the KLFE maybe the second-best merging algorithm. The KLFE is a social- 6. Limitations psychology-based merging algorithm; different from the The selected algorithms are not the most recent; GFM and IDM, it additionally considers fuel efﬁciency in its model. It seems that among social-psychology-based nevertheless, they are very classical and representative that many more advanced algorithms developed their framework merging algorithms, the one that considers more factors can achieve a better inﬂuence on the vehicle platoon on the based on the extension of these classical algorithms. Given mainline. This might be because the merging driver’s that it is hard to evaluate all merging algorithms, the social-psychology desire as well as additional optimal evaluation on classical algorithms would be a more practical targets for the mainline trafﬁcwereboth considered. The way. While that the four algorithms selected in the research Figure 8 Driving comfort level of automatic merging algorithm 43 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 Chou, F.-C., Shladover, S.E. and Bansal, G. (2016), may not be enough to represent the two categories of the merging algorithms, more merging algorithms should be “Coordinated merge control based on V2V communication”, added to evaluate in the future. 2016 IEEE Vehicular Networking Conference (VNC). IEEE, pp. 1-8. Cooper, D.F. and Ferguson, N. (1976), “Trafﬁc studies at T- 7. Conclusions Junctions. 2: A conﬂict simulation record”, Trafﬁc This study analyzed CAV merging behaviors’ inﬂuence on the Engineering & Control, Vol. 17 No. Analytic. mainline vehicle platoon considering driving safety and Dang, D., Gao, F. and Hu, Q. (2020), “Motion planning for comfort, by considering the merging algorithm heterogeneity. autonomous vehicles considering longitudinal and lateral Four types of merging algorithms, including the reference- dynamics coupling”, Applied Sciences, Vol. 10 No. 9, p. 3180. trajectory-based merging algorithm and the social-psychology- Ding, J., Li, L., Peng, H. and Zhang, Y. (2019), “A rule-based based merging algorithm, were evaluated by using a self- cooperative merging strategy for connected and automated developed multi-driver simulator system. The results show that vehicles”, IEEE Transactions on Intelligent Transportation although these algorithms achieved good performance in their Systems, Vol. 21 No. 8, pp. 3436-3446. original simulation studies, their effects were quite various Fellendorf, M. and Vortisch, P. (2001), “Validation of the when the realistic driving behavior was embedded into the microscopic trafﬁc ﬂow model VISSIM in different real- experiment. The results show that the reference-trajectory- world situations”, Transportation Research Board 80th Annual based merging algorithm may outperform the social- Meeting. psychology-based merging algorithm which only considers the Hayward, J.C. (1972), “Near miss determination through use ramp vehicle. More types of CAV merging algorithms need to of a scale of danger”. be veriﬁed and compared in the future to form a solid Helbing, D. and Tilch, B. (1998), “Generalized force model of conclusion. trafﬁc dynamics”, Physical Review E, Vol. 58 No. 1, p. 133. Karimi, M., Roncoli, C., Alecsandru, C. and Papageorgiou, M. (2020), “Cooperative merging control via trajectory References optimization in mixed vehicular trafﬁc”, Transportation Ahammed, M.A., Hassan, Y. and Sayed, T.A. (2008), Research Part C: Emerging Technologies, Vol. 116, p. 102663. “Modeling driver behavior and safety on freeway merging Kuutti, S., Bowden, R., Joshi, H., de Temple, R. and Fallah, S. areas”, Journal of Transportation Engineering, Vol. 134 No. 9, (2019), “End-to-end reinforcement learning for pp. 370-377. autonomous longitudinal control using advantage actor critic Aksjonov, A., Ricciardi, V., Augsburg, K., Vodovozov, V. and with temporal context”, 2019 IEEE Intelligent Transportation Petlenkov, E. (2020), “Hardware-in-the-loop test of an Systems Conference (ITSC). IEEE, pp. 2456-2462. open-loop fuzzy control method for decoupled Letter, C. and Elefteriadou, L. (2017), “Efﬁcient control of electrohydraulic antilock braking system”, IEEE Transactions fully automated connected vehicles at freeway merge on Fuzzy Systems, Vol. 29 No. 5, pp. 965-975. segments”, Transportation Research Part C: Emerging Artunedo, A., Godoy, J., Haber, R., Villagra, J. and del, T.,R. Technologies, Vol. 80, pp. 190-205. M. (2015), “Advanced co-simulation framework for Lu, X.-Y. and Hedrick, J.K. (2003), “Longitudinal control cooperative maneuvers among vehicles”, 2015 IEEE 18th algorithm for automated vehicle merging”, International International Conference on Intelligent Transportation Systems. Journal of Control, Vol. 76 No. 2, pp. 193-202. IEEE, pp. 1436-1441. Lu, X.-Y., Tan, H.-S., Shladover, S.E. and Hedrick, J.K. Awal, T., Kulik, L. and Ramamohanrao, K. (2013a), “Ak- (2004), “Automated vehicle merging maneuver leader fuel-efﬁcient trafﬁc model”, 2013 IEEE Intelligent implementation for AHS”, Vehicle System Dynamics, Vol. 41 Vehicles Symposium (IV). IEEE, pp. 1014-1021. No. 2, pp. 85-107. Awal, T., Kulik, L. and Ramamohanrao, K. (2013b), “Optimal McCartt, A.T., Northrup, V.S. and Retting, R.A. (2004), trafﬁc merging strategy for communication-and sensor- “Types and characteristics of ramp-related motor vehicle enabled vehicles”, 16th International IEEE Conference on crashes on urban interstate roadways in Northern Virginia”, Intelligent Transportation Systems (ITSC 2013). IEEE, Journal of Safety Research, Vol. 35 No. 1, pp. 107-114. pp. 1468-1474. Madhusudhanan, A.K. (2019), “A method to improve an Bellem, H., Schönenberg, T., Krems, J.F. and Schrauf, M. electric vehicle’s range: efﬁcient cruise control”, European (2016), “Objective metrics of comfort: developing a driving Journal of Control, Vol. 48, pp. 83-96. style for highly automated vehicles”, Transportation Research Mohammed, H., Bigazzi, A.Y. and Sayed, T. (2019), Part F: Trafﬁc Psychology and Behaviour, Vol. 41, pp. 45-54. “Characterization of bicycle following and overtaking Bellem, H., Thiel, B., Schrauf, M. and Krems, J.F. (2018), maneuvers on cycling paths”, Transportation Research Part C: “Comfort in automated driving: an analysis of preferences Emerging Technologies, Vol. 98, pp. 139-151. for different automated driving styles and their dependence Nalic, D., Pandurevic, A., Eichberger, A. and Rogic, B. on personality traits”, Transportation Research Part F: trafﬁc (2020b), “Design and implementation of a co-simulation Psychology and Behaviour, Vol. 55, pp. 90-100. framework for testing of automated driving systems”, Bieker-Walz, L., Behrisch, M. Junghans, M. and Gimm, K. Sustainability, Vol. 12 No. 24, p. 10476. (2017), “Evaluation of car-following-models at controlled Nalic, D., Li, H., Eichberger, A., Wellershaus, C., Pandurevic, intersections”. A. and Rogic, B. (2020a), “Stress testing method for 44 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 scenario-based testing of automated driving systems”, IEEE using Gaussian ﬁnite mixture models”, The R Journal, Vol. 8 Access, Vol. 8, pp. 224974-224984. No. 1, p. 289. Nandi, A.K., Chakraborty, D. and Vaz, W. (2015), “Design of Sefati, M., Chandiramani, J., Kreisköther, K., Kampker, A. a comfortable optimal driving strategy for electric vehicles and Baldi, S. (2017), “Towards tactical behaviour planning using multi-objective optimization”, Journal of Power Sources, under uncertainties for automated vehicles in urban Vol. 283, pp. 1-18. scenarios”, 2017 IEEE 20th International Conference on Nguyen, N.A., Moser, D., Schrangl, P., del Re, L. and Jones, Intelligent Transportation Systems (ITSC). IEEE, pp. 1-7. S. (2017), “Autonomous overtaking using stochastic model Treiber, M., Hennecke, A. and Helbing, D. (2000), predictive control”, 2017 11th Asian Control Conference “Congested trafﬁc states in empirical observations and (ASCC). IEEE, pp. 1005-1010. microscopic simulations”, Physical Review E,Vol.62 No. 2, Olofsson, M. and Pettersson, J. (2015), Parameterization and p. 1805. Validation of Road and Driver Behavior Models for Carmaker Wang, Y., Wenjuan, E., Tang, W., Tian, D., Lu, G. and Yu, G. Simulations and Transmission HIL-Rig. (2013), “Automated on-ramp merging control algorithm Potzy, J., Feinauer, S., Siedersberger, K.-H. and Bengler, K. based on internet-connected vehicles”, IET Intelligent (2019), “Manual drivers’ evaluation of automated merging Transport Systems, Vol. 7 No. 4, pp. 371-379. behavior in dense trafﬁc: efﬁciency matters”, 2019 IEEE Weng, J. and Meng, Q. (2014), “Rear-end crash potential Intelligent Transportation Systems Conference (ITSC). IEEE, estimation in the work zone merging areas”, Journal of pp. 3454-3460. Advanced Transportation, Vol. 48 No. 3, pp. 238-249. Reinolsmann, N., Alhajyaseen, W., Brijs, T., Pirdavani, A., Weng, J., Xue, S., Yang, Y., Yan, X. and Qu, X. (2015), “In- Hussain, Q. and Brijs, K. (2019), “Investigating the impact depth analysis of drivers’ merging behavior and rear-end of dynamic merge control strategies on driving behavior on crash risks in work zone merging areas”, Accident Analysis & rural and urban expressways – a driving simulator study”, Prevention, Vol. 77, pp. 51-61. Transportation Research Part F: Trafﬁc Psychology and Wu, P., Gao, F. and Li, K. (2019), “A vehicle type dependent Behaviour, Vol. 65, pp. 469-484. car-following model based on naturalistic driving study”, Samiee, S.,Azadi,S., Kazemi, R.and Eichberger,A. (2016), Electronics, Vol. 8 No. 4, p. 453. “Towards a decision-making algorithm for automatic lane Yue, L., Abdel-Aty, M., Wu, Y., Yuan, J. and Morris, M. change manoeuvre considering trafﬁcdynamics”, PROMET – (2020), “Inﬂuence of pedestrian-to-vehicle technology on Trafﬁc&Transportation,Vol.28No.2,pp.91-103. drivers’ response and safety beneﬁts considering pre-crash Schreiber, V., Ivanov, V., Augsburg, K., Noack, M., Shyrokau, conditions”, Transportation Research Part F: Trafﬁc Psychology B., Sandu, C. and Els, P.S. (2018), “Shared and distributed and Behaviour, Vol. 73, pp. 50-65. X-in-the-loop tests for automotive systems: feasibility study”, IEEE Access, Vol. 6, pp. 4017-4026. Corresponding author Scrucca, L., Fop, M., Murphy, T.B. and Raftery, A.E. (2016), Lishengsa Yue can be contacted at: 2017lishengsa@knights. “Mclust 5: clustering, classiﬁcation and density estimation ucf.edu For instructions on how to order reprints of this article, please visit our website: www.emeraldgrouppublishing.com/licensing/reprints.htm Or contact us for further details: permissions@emeraldinsight.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Intelligent and Connected Vehicles Emerald Publishing http://www.deepdyve.com/lp/emerald-publishing/effects-of-connected-and-autonomous-vehicle-merging-behavior-on-Y6KXxvzGGN

Loading next page...

Publisher: Emerald Publishing
Copyright: © Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang.
ISSN: 2399-9802
DOI: 10.1108/jicv-08-2021-0013
Publisher site: See Article on Publisher Site

Abstract

Purpose – This study aims to evaluate the inﬂuence of connected and autonomous vehicle (CAV) merging algorithms on the driver behavior of human-driven vehicles on the mainline. Design/methodology/approach – Previous studies designed their merging algorithms mostly based on either the simulation or the restricted ﬁeld testing, which lacks consideration of realistic driving behaviors in the merging scenario. This study developed a multi-driver simulator system to embed realistic driving behavior in the validation of merging algorithms. Findings – Four types of CAV merging algorithms were evaluated regarding their inﬂuences on driving safety and driving comfort of the mainline vehicle platoon. The results revealed signiﬁcant variation of the algorithm inﬂuences. Speciﬁcally, the results show that the reference-trajectory-based merging algorithm may outperform the social-psychology-based merging algorithm which only considers the ramp vehicles. Originality/value – To the best of the authors’ knowledge, this is the ﬁrst time to evaluate a CAV control algorithm considering realistic driver interactions rather than by the simulation. To achieve the research purpose, a novel multi-driver driving simulator was developed, which enables multi-drivers to simultaneously interact with each other during a virtual driving test. The results are expected to have practical implications for further improvement of the CAV merging algorithm. Keywords Driving simulator, Connected and autonomous vehicle, Merging algorithm, Merging behavior, Safety and comfort, Driving safety, Driving comfort Paper type Research paper merging path generation algorithm and social-psychology- 1. Introduction based merging algorithm. For the ﬁrst type, the objective of Connected and autonomous vehicle (CAV) technology has this type of merging control is to ﬁnd a reference trajectory been gaining more and more attention in recent years; it to guide the autonomous vehicle by considering physical releases drivers from heavy driving tasks and avoids driver and kinematic restrictions to fulﬁll a successful merging. Lu errors. One challenge of CAV technology is its adaptability and Hedrick (2003) and Lu et al.(2004) designed a set of in critical trafﬁc scenarios. A typical critical scenario is the kinematical restriction functions in terms of the acceleration merging scenario at the freeway ramp area; it is the hotspot and position, to ensure that the vehicles can reach the of trafﬁc crashes. In total, 18% of all interstate freeway merging pointsatan appropriate time. Wang et al. (2013) crashes, 17% of the injury crashes and 11% of the fatal proposed a cooperative driving algorithm based on vehicular crashes occurred at interchanges, and most proportion of operation characteristics for the ramp merging. They these crashes took place at the entrance or exit ramps considered the position and speed requirements for both (Ahammed et al., 2008; McCartt et al.,2004). Given that one and two vehicles on the ramp. The second type there is usually signiﬁcant vehicle interaction at the merging considers driving preference regarding the merging behavior areas, the design of the CAV merging algorithm is critically important; the algorithm is supposed to ensure a safe © Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang. Published in merging behavior; meanwhile, it is expected to disturb the Journal of Intelligent and Connected Vehicles. Published by Emerald mainline driving as little as possible. Publishing Limited. This article is published under the Creative Commons Many studies have designed CAV merging control Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, algorithms. Generally, the merging algorithms can be translate and create derivative works of this article (for both commercial categorized into two types: physical-restriction-based and non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence maybe seen at http:// creativecommons.org/licences/by/4.0/legalcode The current issue and full text archive of this journal is available on Emerald The authors acknowledge the ﬁnancial support of the Innovative Insight at: https://www.emerald.com/insight/2399-9802.htm Technology Administration of US Department of Transportation, Award No. DTRT13-G-UTC53 (SAFER-SIM). Received 19 August 2021 Journal of Intelligent and Connected Vehicles Revised 18 October 2021 5/1 (2022) 36–45 8 November 2021 Emerald Publishing Limited [ISSN 2399-9802] [DOI 10.1108/JICV-08-2021-0013] Accepted 23 November 2021 36 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 and manipulates the merging model based on factors such as CarMaker, TFSS and Matlab/SIMULINK is often used desired gap distance and desired mainstream speed. (Madhusudhanan, 2019; Nalic et al., 2020a; Nalic et al., Basically, this type of algorithm deﬁnes the model in a form 2020b). of preferred and actual accelerations and distance gap However, the lack of realistic driver behavior in the algorithm maintain (Awal et al., 2013a, 2013b; Chou et al., 2016; validation deteriorates the credibility of the algorithm. As Karimi et al., 2020; Wu et al., 2019). An advanced form of pointed by Aksjonov et al. (2020), purely computation thesecond typeis to consider multipleoptimization targets simulation does not guarantee realistic environments for a and generate the reference merging path by solving the testing vehicle, and this is the reason that in recent years the optimal solution. Ding et al. (2019) proposed a rule-based concept of “hardware-in-the-loop” or “human-in-the-loop” merging algorithm with minimizing travel time and delay as becomes more prevalent (Aksjonov et al., 2020; Artunedo et al., the target; they formed a closed-form analytical solution to 2015; Dang et al., 2020; Schreiber et al.,2018). Basically, using real hardware or a driver to test the algorithm is more reliable achieve a near-optimal merging sequence. Letter and Elefteriadou (2017) presented a longitudinal freeway than using a driver behavior model particularly when it is merging control algorithm, which used the average travel necessary to observe vehicle interactions and possible improper speed as the optimization target, and they used LINGO to driving adaptation behaviors such as driving errors and resolve the optimal solution. aggressive driving. Theoretically speaking, a driver behavior The veriﬁcation of the above algorithms is usually based on model is controlled by many kinematical restrictions with the simulation platforms, either through a single platform or a co- purpose of generating “smooth” behaviors, and it is hard to simulation based integrated simulation platform. Some single mimic improper driving adaptations (Nalic et al., 2020a). platforms are basically microscopic trafﬁc ﬂow simulation Regarding the CAV merging control algorithms, most of the software (TFSS) such as Vissim and SUMO; they have aforementioned studies were based on simulation, which is embedded driving behavior models, which can mimic car- hard to represent realistic driver behaviors in the merging scenario. Very few of them conducted ﬁeld testing; however, following or lane change behaviors. For example, Vissim used a rule-based algorithm to initiate lateral lane change behavior and because of the safety consideration, only conservative a psychophysical model for the longitudinal car-following algorithms and restricted testing conditions (such as low movement (Fellendorf and Vortisch, 2001). Different from driving speed) were tested. Many studies proved that driver Vissim, SUMO used a Krauss model as its default car- behavior can signiﬁcantly affect the crash and safety level at the following model (Bieker-Walz et al.,2017). However, merging area. Weng et al. (2015) found that the drivers’ individual vehicle dynamics modeling is not a strength for these merging behavior is highly correlated with the rear-end crash platforms. On the contrary, some other single platforms, such risk; there will be high rear-end crash risks when the merging as CarMaker, have a better ability of modeling vehicle dynamic vehicle travels at either a very high or low speed. Weng and details such as powertrain and sensor system. The driver Meng (2014) found that if the merging action initiates earlier, behavior model used in the CarMaker is based on a there will be a lower rear-end crash potential. Reinolsmann proportional-integral-derivative controller, with considerations et al. (2019) also suggested earlier lane change because it can of psychological studies and measurements from real test contribute to smooth maneuvers and gradual speed reductions drivers (Olofsson and Pettersson, 2015). Whatever driver particularly at the rural expressway ramp area. Moreover, Potzy behavior models they are, the models result from speed, speed et al. (2019) concluded that drivers on the mainline prefer an difference, distance gap, vehicle dynamic restrictions or efﬁcient lane change of the autonomous vehicle on the ramp, individual driver characteristics. and results show that drivers would tolerate less compliance Integrated simulation platforms provide more explicit with safety distance to have less interacting trafﬁc. It is quite simulation regarding individual vehicle dynamics or a better necessary to consider the realistic driver behavior for CAV power to optimize CAV control algorithms during the merging control algorithms so that the CAV merging behavior running time; the integrated platform can consist of a TFSS can be more acceptable and predictable for mainline drivers. and several other simulators such as IPG CarMaker and Therefore, the objective of this study is to evaluate the Matlab/SIMULINK. Basically, the TFSS is more efﬁcient inﬂuence of CAV merging algorithms on driver behavior of at simulating microscopic trafﬁc ﬂow while it ignores the human-driven vehicles on the mainline, by using the human-in- vehicle dynamic details; on the contrary, the IPG CarMaker the-loop concept. Several classical merging algorithms were can better simulate the vehicle dynamics such as the tested in this study, representing the physical-restriction-based powertrain system and sensor system; while the Matlab/ merging path generation algorithm and social-psychology- SIMULINK is mainly for algorithm optimization purpose, based merging algorithm; then their inﬂuence on the mainline its embedded mathematical toolboxes can be used to resolve trafﬁc was analyzed given their algorithm framework features. optimization problems and generate optimized parameters/ This study is expected to conclude design principles of merging outputs of control algorithms. When complicated trafﬁc algorithms that have less inﬂuence on the mainline trafﬁc. The ﬂow simulation is not necessary, IPG CarMaker is often driving safety and driving comfort of mainline human drivers used to verify CAV control algorithms, such as longitudinal would be analyzed to distinguish the performance of different cruise control (Kuutti et al., 2019), lateral lane change merging algorithms. To account for crash risks in a realistic (Samiee et al., 2016), overtaking path planning (Nguyen ﬁeld testing, a driving simulator experiment would be used et al., 2017) and tactical behavior planning (Sefati et al., instead. 2017); while the CAV algorithm is required to be tested in This study is organized as follows: CAV Merging Algorithm certain trafﬁc ﬂow conditions, a co-simulation between section 2 introduces classical CAV merging algorithms that 37 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 were tested in this study; Experimental Design section 3 based on the optimization model was not investigated in this presents the multi-driver simulator system developed for this study, because of its computation cost in the real-time driving study, experimental scenarios and data analysis method; simulator system. These tested merging algorithms and their Results section 4 presents the results and Discussion section 5 examples are listed below. investigates the results; ﬁnally, the Limitation section 6 Reference-trajectory-based merging algorithm. This study presents the study limitation and Conclusion section 7 adopted the algorithm (denoted as AHS) and its parameters proposed by Lu and Hedrick (2003) and Lu et al. (2004) as an summarizes this study. example, which deﬁned the reference trajectory v (t)as md follows: 2. Connected and autonomous vehicle merging algorithm v t ¼ vt ; > mdðÞ merg ðÞ merg Almost all CAV merging algorithm designs adopt a concept of < 1 a t vt 1 a t vðÞ t 1 ðÞ ðÞ i ðÞ merg ðÞ i p i v ðÞ t ¼ when t ðÞ t 1 iDt T ; (1) “virtual platoon,” with a connotation of projecting the CAV’s md i merg merg virt vðÞ t 1 position from the merging ramp to the mainline, and generating p i when T <ðÞ t 1 iDt T ; virt merg merg a “virtual platoon” consists of both human-driven vehicles (yellow) and projected CAV (gray) (Figure 1). The projection is exactly based on geometric parameters of the CAV, and it a t ¼ a t ; b > 0 (2) ðÞ ðÞ i i determines the relative position of the projection vehicle to other mainline vehicles. The core idea of the CAV merging algorithm is to manipulate the speed and acceleration of the v t 1 Dt ðÞ p j projected CAV, so that it can maintain a safe headway distance j¼1 a t ¼ (3) 0ðÞ i X to the front vehicle under a desired traveling speed. Basically, in vðÞ t 1 Dt1 dist para p j a fully connected and automated environment, a central j¼1 controller will be set up covering the upstream and downstream of merging area, and collect speed and location information of where v(t) is the merging vehicle speed, v (t) is the speed of the all vehicles (both on mainline and ramp) entering the control ﬁrst vehicle in the platoon on mainline, t is the time when merg area; the vehicles in the control area will be manipulated so that the merging algorithm starts, T is the time when the virtual virt the speed and headway distance of each single vehicle in the platoon is established but merging is not complete yet, T is merg virtual platoon can be well accommodated. Speciﬁcally, the time when the merging is ﬁnished, b is a coefﬁcient and the controller will accommodate the autonomous vehicle dist_para is the initial distance relationship between vehicles (projected) based on its relative speed and position to the considering desired distance before and after the merging. leading vehicle (ﬁrst vehicle in the platoon), and consecutively Detailed variable deﬁnitions can be found in Lu and Hedrick accommodate the second and third vehicles based on similar (2003) and Lu et al.(2004). safety considerations toward the vehicles in front of them. Social-psychology-based merging algorithm. This study tested In this study, the driving environment is partially connected, three examples, which were borrowed from car-following and only the CAV can manage its movement by sensing the models, by considering the driver’s desire to main a certain leading vehicle; for the second and third vehicles, drivers need to speed and distance to the leading vehicle. The intelligent driver determine the driving by themselves rather than follow the central model (IDM) (Treiber et al.,2000), the generalized force controller. Therefore, the controller, which is embedded with the model (GFM) (Helbing and Tilch, 1998) and the k-leader fuel- merging algorithm, will take the information of the ﬁrst vehicle as efﬁcient (KLFE) model (Awal et al., 2013a, 2013b) were used input to manage the movement of the CAV. as examples in this study. The model parameters were identical Two types of classical CAV merging algorithms, the to the ones in corresponding studies. reference-trajectory-based merging algorithm generation The IDM is given by: algorithm and the social-psychology-based merging algorithm, "# were reproduced in this study based on previous studies. The v S ðÞ v;Dv desire a ¼ A 1 (4) study veriﬁed their effects on the vehicle platoon on the V S mainline. The third type of CAV merging algorithm that is rﬃﬃﬃﬃﬃ v vDv Figure 1 Virtual platoon and projected CAV S ðÞ v; Dv ¼ g 1 g 1 vT 1 pﬃﬃﬃﬃﬃﬃ (5) desire 0 1 2 Ab where a is the suggested acceleration; v and V are the current speed and desired speed, respectively; Dv is the speed difference to the preceding vehicle; S and S are the current desire following distance and desired following distance, respectively; A and b are maximum desired acceleration and deceleration, respectively; g and g are different jam distance parameters 0 1 and d is a constant coefﬁcient. Detailed variable deﬁnitions can be found in Treiber et al.(2000). The GFM is given by: 38 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 sns ðÞ v desire the preceding vehicle; T is the safe time headway; S is the jam Ra V 1 e v a ¼ b (6) distance. Detailed variable deﬁnitions can be found in Awal et al.(2013a, 2013b). It is worth mentioning that the KLFE considers vehicles on both the ramp and mainline. s s ðÞ v desire ðÞ d u Dv Dve b ¼ (7) 3. Experimental design 3.1 Apparatus A multi-driver driving simulator system (Figure 2)wasdeveloped S ðÞ v ¼ s 1 vT (8) desire 0 to test a vehicle platoon in a virtual driving scenario. Compared with a realistic ﬁeld test, the advantage of using the driving where a is the nth vehicle’s acceleration; v and V are the current simulator isthatitcan test dangerousdriving scenarioswithout speed and desired speed, respectively; Dv is the speed difference real collision risks. The simulator system designed a data to the preceding vehicle; s , s and s are the current following n desire 0 collection module, a vehicle physics module, a scenario distance, desired safe following distance and minimum following management module and a communication module. The data distance, respectively; u is the Heaviside function; T is the safe collection module collects driver behavior data in the scenario, time headway; t and t are the acceleration time and braking a d such as brake, throttle, steering wheel, speed and position; the time, respectively; R and R arethe rangeofthe acceleration and a d vehicle physics module simulates vehicle dynamics and related range of the braking interaction, respectively. Detailed variable physical features, such as engine dynamics and collision effects; deﬁnitions can be found in Helbing and Tilch(1998). the scenario management module conﬁgures scenario control The KLFE is given by: scripts, and it manages all types of scenario objects and their ðÞ ðÞ v t1Dt ¼ min v t1Dt (9) actions; as for the communication module, it connects multiple n n;m driver clients and distributes the simulation data between clients simultaneously (Figure 3). a safe ðÞ v t1 Dt ¼ max 0; min v ðÞ t ; v ðÞ t (10) n;m n;m n;m 3.2 Merging scenario design This study designed a merging scenario as illustrated in v ðÞ t ¼ v ðÞ t 1 k (11) n;m Figure 4. Three human-driven vehicles (the ﬁrst to third yellow vehicles) are traveling on the mainline, and they form a stable 2 3 4 vehicle platoon. A CAV (fourth red vehicle) is merging into the desire v ðÞ t s ðÞ t v ðÞ t n n n;m 4 5 mainline from a ramp, and it is supposed to cut in between the k ¼ ADt 1 (12) V V n;m ﬁrst and second vehicles in the platoon; the CAV is controlled by the automatic merging algorithm. The second human- driven vehicle determines whether to yield to the CAV, based hi 2 0 v ðÞ t p ¼ 2 x ðÞ t x ðÞ t l v ðÞ t Dt (13) on the safety consideration. Normally, the CAV would appear m n n ahead in the second human-driven vehicle’s view; therefore, the second human-driven vehicle would slow down. However, in bDt1qif q 0 safe rare cases, the second human-driven vehicle decides to v ðÞ t ¼ (14) n;m bDt1 v ðÞ t if q < 0 accelerate and overtake the CAV; in these cases, the merging algorithm will recognize that the second human-driven vehicle qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ would arrive at the merging point before the CAV, thus the ðÞ q ¼ b Dt bp (15) algorithm will change its goal to following the second human- driven vehicle. The study arranged a ﬁfth environmental vehicle (blue) in front of the vehicle platoon in the mainline; the desire s ðÞ t ¼ S 1 b ðÞ n m (16) n;m vehicle follows a predeﬁned path and speed, and it is used as a v ðÞ t v ðÞ t v ðÞ t n n m Figure 2 Multi-driver driving simulator system framework b ¼ max 0; v ðÞ t T 1 pﬃﬃﬃﬃﬃﬃﬃﬃﬃ (17) 2 Ajbj where n and m stand for the nth and mth vehicles in the platoon; desire s is the desired following distance between the nth and mth n;m vehicles; v is the following vehicle speed when the distance n;m between the following and preceding vehicle is large, whereas safe the v is the following vehicle safe speed when the gap distance n;m is small; v , v and V are the current following speed, current n m preceding speed and desired speed, respectively; x and x are n m the following and preceding vehicle positions; l is the effective size plus a margin; A is the maximum desired acceleration; b is the maximum braking rate; b is the estimated braking rate of 39 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 group of drivers was given a practice driving so that they can be Figure 3 Multi-driver driving simulator system and experiment familiar with the driving environment. During the formal experiment, each group drove four tracks, and each track had one type of merging algorithm repeated six times. In other words, the drivers experienced ﬁve merging scenarios in one track. 3.4 Inﬂuence period during merging It is worth mentioning that during the above “merging scenario,” the CAV and second vehicle would not interact with each other all the time. Because of the sight of the second driver’sview, only when the CAV is close enough, the second driver would be affected and accommodate driving behaviors to the merging vehicle. This period is deﬁned as the inﬂuence period. Given that the driving behaviors are assumed to be different between the normal and inﬂuenced driving periods, the ﬁnite Gaussian mixture model (GMM) was used to distinguish two periods in the merging scenario. The GMM assumes that data of different features is coming from a mixture of two or more Gaussian distributions (i.e. clusters), and the GMM allocates data points to most probable distributions by expectation- maximization (EM) algorithm (Scrucca et al., 2016). Similar trajectory clustering practice was conducted by Mohammed et al.(2019);the researchers used ﬁnite GMM to cluster cyclists overtaking and following trajectories into different states. In the merging scenario, the second driver manages the throttle and brake to maintain the safety buffer between both the ﬁrst vehicle and CAV. Therefore, this study used second vehicle’s speed, throttle and brake positions and distances to reference vehicle to control the speed of the vehicle platoon, in the ﬁrst vehicle and CAV as trajectory features to be clustered. case either too fast or too slow. Figure 4 shows an example of the throttle clusters generated by In each experiment, a set of three connected drivers drove GMM. It shows that at the time of around 100, the second through a track that contains ﬁve merging scenarios. These ﬁve driver notices the merging CAV and begins to monitor the scenarios have the same merging algorithms, and the driver’s collision risk; then at the time of around 160, the second driver average driving performance was analyzed later. The begins to release the pedal position to slow down; during the experiment was a within-subjects experiment, and three drivers time point of 100–160, the throttle position does not change, experienced all ﬁve merging algorithms; therefore, in total, ﬁve this might be because of the driver’s reaction time delay. It is experiments were conducted. The merging algorithms were worth mentioning that the driving periods of the second driver presented in a randomized way to account for the order effect were applied to the third driver in this study (Figure 5). (Yue et al., 2020). In the experiment, the distance to the front vehicle and its speed information was displayed on the following vehicle’s screen; referring to this information display, 3.5 Driving performance metrics each driver was asked to follow the front vehicle keeping a This study mainly investigated the inﬂuence of CAV merging distance of 60–80 m; the speed limit was 55 mph. algorithms on the second and third human-driven vehicles. Two driving periods were analyzed: the merging and following 3.3 Experimental procedure periods. The merging period is deﬁned as the period from the In total, 16 groups of drivers (i.e. 48 drivers) conducted the experiment. They were driving on the mainline and an Figure 5 Throttle position during merging scenario autonomous car controlled by the merging algorithm was merging from the ramp. Before the formal experiment, each Figure 4 Merging scenario in experiment 40 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 time the CAV begins to move on the ramp to the time it arrives Nandi et al.,2015). In the following period, the minimum at the merging point, which occurs in the merging scenario. headway distance was used to indicate the driving comfort level The following period (Figure 6) is from the time that (Bellem et al.,2016); the larger the headway distance, the more the CAV ﬁnishes merging and begins to drive with the vehicle comfortable the driving is. platoon, to the time that the CAV leaves the lane of vehicle platoon (the leaving point was given); the following actions 4. Results speciﬁcally refer to the actions of second human-driven vehicle 4.1 Inﬂuence on driving safety (following the CAV) and third human-driven vehicle (following During the merging period, for the second driver, both the the second human-driven vehicle). minimum TTC and DRAC between the four merging For each driving period, two aspects of metrics were algorithms were signiﬁcantly different (F = 5.62, p-value < collected in terms of both driving safety and driving comfort. 0.01 and F = 14.45, p-value < 0.01, respectively). Figure 7 The driving safety measures include the minimum time-to- shows that the AHS algorithm had the largest minimum TTC, collision (TTC) and the deceleration rate to avoid crashes which was 19.45 s; signiﬁcantly smaller than the AHS (DRAC). The TTC was introduced by Hayward(1972) and is algorithm, the GFM algorithm had a minimum TTC of only the most widely used surrogate safety measure (SSM); it 6.49 s. As for the IDM and KLFE algorithms, their minimum indicates the time that is left for the following vehicle to hit a TTC values were between AHS and GFM; the KLFE had a leading vehicle. The TTC is given by: slightly higher safety level than the IDM. In addition, the AHS algorithm had a very small DRAC of only 0.007; the KLFE’s S L ; v > v 2 1 DRAC was slightly larger, i.e. 0.009, whereas the GFM and TTC ¼ (18) v v 2 1 IDM had a much larger DRAC of around 0.7. For the third 1; otherwise driver, only the minimum TTC was found signiﬁcantly different between merging algorithms (F = 2.45, p-value = where S is the distance gap between the leading and following 0.056). Similar to the inﬂuence on the second driver, the AHS vehicles; L is the vehicle length; v and v are speeds of 2 1 had the largest minimum TTC, which was 36.77 s; the GFM following and leading vehicles, respectively. TTC can be had the smallest minimum TTC of only 22.16 s. The level of calculable only when v is greater than v . In this study, TTC 2 1 minimum TTC of the IDM and KLFE was between AHS and was assigned a large value of 100 when v < v . Obviously, a 2 1 GFM. large value of minimum TTC indicates a high safety level. During the following period, the second driver’s minimum The DRAC is also a widely used SSM, proposed by Cooper TTC was found signiﬁcantly different between four types of and Ferguson (1976); it indicates the required minimum merging algorithms (F = 2.31, p-value = 0.068). The AHS had deceleration rate for a following vehicle to avoid a crash with a the largest minimum TTC of 41.30 s, whereas the KLFE had leading vehicle. The DRAC is given by: the smallest minimum TTC of 30.23 s. The GFM was slightly 2 larger than KLFE, which was 36.20 s, and the IDM had a v v ðÞ 2 1 ; v > v minimum TTC of 38.40 s. Similar to the second driver, the 2 1 DRAC ¼ (19) S L third driver had signiﬁcantly different minimum TTC between 0; otherwise merging algorithms (F = 3.81, p-value < 0.01). Speciﬁcally, the AHS had the largest minimum TTC of 11.61 s, whereas the Usually, a threshold DRAC would be selected and the GFM and IDM had the smallest minimum TTC of around percentage of DRAC > DRAC is used to indicate the safety 8.35 s. The minimum TTC of KLFE was 9.58 s. level; a large percentage represents a dangerous situation. In this study, a threshold DRAC of 3.0 s was adopted, and the 4.2 Inﬂuence on driving comfort “DRAC” term mentioned in later sections is the percentage During the merging period, the second driver’s mean value, weighted by trajectory length. Both the minimum TTC deceleration was signiﬁcantly different between the four and DRAC were applied to the merging and following periods. merging algorithms (F = 4.74, p-value = 0.002). Figure 8 In terms of the driving comfort measures, the mean shows that the AHS had the smallest deceleration of 2.64 m/s , deceleration/acceleration and the average jerk during the whereas the GFM had the largest deceleration of 4.23 m/s . deceleration/acceleration processes were used for the merging The decelerations of the KLFE and IDM were in between with period. The jerk is the derivative of deceleration/acceleration. In a value of 3.38 and 3.11 m/s , respectively. The average jerks the merging period, these measures were proved to have a negative during the deceleration process were also found signiﬁcantly relationship with driving comfort (Bellem et al., 2016, 2018; different between the merging algorithms (F = 14.45, p-value < 0.01). To be speciﬁc, the GFM had the largest average jerk Figure 6 Following period when a CAV cuts in between the ﬁrst and during the deceleration process, which was 36.86; the IDM and second vehicles AHS had a relatively smaller average jerk value close to each other of around 27.50; the jerk value of the KLFE was in between and it was 30.04. In terms of the minimum headway distance, a signiﬁcant difference was found because of the merging algorithms (F = 62.26, p-value < 0.01). The AHS and KLFE had a very large minimum headway distance of 37.99 and 36.11 m, respectively, whereas the GFM had the smallest 41 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 minimum headway distance of 17.57 m. The IDM had a the other three algorithms are very close. In terms of the minimum headway distance of 23.64 m, slightly larger than the minimum headway distance, the AHS had the largest one of GFM. For the third driver, the average jerks during the 53.59 m, whereas the other three algorithms are very close. acceleration process were found signiﬁcantly different between the merging algorithms (F = 2.21, p-value = 0.08). The AHS 5. Discussions had the largest average jerk of 47.80, whereas the KLFE had The heterogeneity of CAV merging behavior caused the smallest average jerk of 36.83. The average jerks of the signiﬁcantly different inﬂuences on the mainline vehicle GFM and IDM were close to each with a value of 39.62 and platoon. For the second driver, regarding the driving safety in 40.48, respectively. either merging or following periods, the minimum TTC and During the following period, the second driver, both the DRAS show that AHS had the least negative inﬂuence, whereas mean acceleration and deceleration were found to be the GFM had the most negative inﬂuence. Regarding the signiﬁcantly affected by the merging algorithm. For mean driving comfort, the deceleration, jerk and headway distance acceleration, the GFM had the largest one of 4.33 m/s , show that the AHS had the best driving comfort, whereas the whereas the AHS had the smallest one of 3.34 m/s .The GFM had the worst driving comfort in both two driving mean accelerations of IDM and KLFE were 3.79 and periods. The inﬂuence of IDM and KLFE on driving safety and 3.77 m/s , respectively. Additionally, the GFM had the comfort was generally in between the AHS and GFM. largest mean deceleration of 3.47 m/s , whereas AHS had Compared with the second driver, for the third driver, the the smallest mean deceleration of 2.89 m/s .The mean heterogeneity of inﬂuence of CAV merging behaviors deceleration of the IDM and KLFE was close to each other. was much less signiﬁcant. In terms of driving safety, the DARC In terms of the average jerk, it was found to be signiﬁcantly was not signiﬁcantly different and only the minimum TTC was affected by the merging algorithms during the acceleration observed to be varied between CAV algorithms. The minimum process (F = 4.92, p-value < 0.01). The GFM had the TTC shows that in both merging and following periods, the largest one of 46.81, whereas the AHS had the smallest one AHS was the safest, whereas the GFM was the riskiest one. In of 41.93. The average jerks of IDM and KLFE were 43.29 terms of driving comfort, in the merging period, the average and 43.72 m/s , respectively. For the minimum headway jerk shows that the AHS was the least comfortable one, and distance, a signiﬁcant difference was also found (F = 47.21, both the GFM and IDM had a better comfortable level than the p-value < 0.01). The GFM had the smallest one of 21.60 m, whereas the AHS had the largest one of 41.01 m. The KLFE AHS; the KLFE was the most comfortable one. In the also had a large minimum headway distance of 40.69 m, following period, the mean acceleration shows that the GFM was the least comfortable one, whereas the minimum headway whereas the value of IDM was 32.07 m. distance shows that the AHS is the most comfortable one. During the following period, the third driver’s mean The results show that the AHS can guarantee driving acceleration and minimum headway distance were found to be safety and driving comfort among the tested four types of signiﬁcantly affected by the merging algorithm (F = 2.43, p- value = 0.06 and F = 4.11, p-value < 0.01, respectively). The merging algorithms; nevertheless, its driving comfort may GFM had the largest mean acceleration of 4.51 m/s , whereas deteriorate for the later part of the vehicle platoon in the Figure 7 Driving safety level of automatic merging algorithm 42 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 merging period. It was interesting that the AHS, as a GFM was the worst merging algorithm because it caused reference-trajectory-based merging algorithm, can achieve more negative effects on driving safety and driving a comprehensive good performance in terms of both comfort; this might be because it is not much suitable to driving safety and comfort, without considering many capture a driver’s comprehensive driving preference. other optimization goals. In addition, the KLFE maybe the second-best merging algorithm. The KLFE is a social- 6. Limitations psychology-based merging algorithm; different from the The selected algorithms are not the most recent; GFM and IDM, it additionally considers fuel efﬁciency in its model. It seems that among social-psychology-based nevertheless, they are very classical and representative that many more advanced algorithms developed their framework merging algorithms, the one that considers more factors can achieve a better inﬂuence on the vehicle platoon on the based on the extension of these classical algorithms. Given mainline. This might be because the merging driver’s that it is hard to evaluate all merging algorithms, the social-psychology desire as well as additional optimal evaluation on classical algorithms would be a more practical targets for the mainline trafﬁcwereboth considered. The way. While that the four algorithms selected in the research Figure 8 Driving comfort level of automatic merging algorithm 43 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 Chou, F.-C., Shladover, S.E. and Bansal, G. (2016), may not be enough to represent the two categories of the merging algorithms, more merging algorithms should be “Coordinated merge control based on V2V communication”, added to evaluate in the future. 2016 IEEE Vehicular Networking Conference (VNC). IEEE, pp. 1-8. Cooper, D.F. and Ferguson, N. (1976), “Trafﬁc studies at T- 7. Conclusions Junctions. 2: A conﬂict simulation record”, Trafﬁc This study analyzed CAV merging behaviors’ inﬂuence on the Engineering & Control, Vol. 17 No. Analytic. mainline vehicle platoon considering driving safety and Dang, D., Gao, F. and Hu, Q. (2020), “Motion planning for comfort, by considering the merging algorithm heterogeneity. autonomous vehicles considering longitudinal and lateral Four types of merging algorithms, including the reference- dynamics coupling”, Applied Sciences, Vol. 10 No. 9, p. 3180. trajectory-based merging algorithm and the social-psychology- Ding, J., Li, L., Peng, H. and Zhang, Y. (2019), “A rule-based based merging algorithm, were evaluated by using a self- cooperative merging strategy for connected and automated developed multi-driver simulator system. The results show that vehicles”, IEEE Transactions on Intelligent Transportation although these algorithms achieved good performance in their Systems, Vol. 21 No. 8, pp. 3436-3446. original simulation studies, their effects were quite various Fellendorf, M. and Vortisch, P. (2001), “Validation of the when the realistic driving behavior was embedded into the microscopic trafﬁc ﬂow model VISSIM in different real- experiment. The results show that the reference-trajectory- world situations”, Transportation Research Board 80th Annual based merging algorithm may outperform the social- Meeting. psychology-based merging algorithm which only considers the Hayward, J.C. (1972), “Near miss determination through use ramp vehicle. More types of CAV merging algorithms need to of a scale of danger”. be veriﬁed and compared in the future to form a solid Helbing, D. and Tilch, B. (1998), “Generalized force model of conclusion. trafﬁc dynamics”, Physical Review E, Vol. 58 No. 1, p. 133. Karimi, M., Roncoli, C., Alecsandru, C. and Papageorgiou, M. (2020), “Cooperative merging control via trajectory References optimization in mixed vehicular trafﬁc”, Transportation Ahammed, M.A., Hassan, Y. and Sayed, T.A. (2008), Research Part C: Emerging Technologies, Vol. 116, p. 102663. “Modeling driver behavior and safety on freeway merging Kuutti, S., Bowden, R., Joshi, H., de Temple, R. and Fallah, S. areas”, Journal of Transportation Engineering, Vol. 134 No. 9, (2019), “End-to-end reinforcement learning for pp. 370-377. autonomous longitudinal control using advantage actor critic Aksjonov, A., Ricciardi, V., Augsburg, K., Vodovozov, V. and with temporal context”, 2019 IEEE Intelligent Transportation Petlenkov, E. (2020), “Hardware-in-the-loop test of an Systems Conference (ITSC). IEEE, pp. 2456-2462. open-loop fuzzy control method for decoupled Letter, C. and Elefteriadou, L. (2017), “Efﬁcient control of electrohydraulic antilock braking system”, IEEE Transactions fully automated connected vehicles at freeway merge on Fuzzy Systems, Vol. 29 No. 5, pp. 965-975. segments”, Transportation Research Part C: Emerging Artunedo, A., Godoy, J., Haber, R., Villagra, J. and del, T.,R. Technologies, Vol. 80, pp. 190-205. M. (2015), “Advanced co-simulation framework for Lu, X.-Y. and Hedrick, J.K. (2003), “Longitudinal control cooperative maneuvers among vehicles”, 2015 IEEE 18th algorithm for automated vehicle merging”, International International Conference on Intelligent Transportation Systems. Journal of Control, Vol. 76 No. 2, pp. 193-202. IEEE, pp. 1436-1441. Lu, X.-Y., Tan, H.-S., Shladover, S.E. and Hedrick, J.K. Awal, T., Kulik, L. and Ramamohanrao, K. (2013a), “Ak- (2004), “Automated vehicle merging maneuver leader fuel-efﬁcient trafﬁc model”, 2013 IEEE Intelligent implementation for AHS”, Vehicle System Dynamics, Vol. 41 Vehicles Symposium (IV). IEEE, pp. 1014-1021. No. 2, pp. 85-107. Awal, T., Kulik, L. and Ramamohanrao, K. (2013b), “Optimal McCartt, A.T., Northrup, V.S. and Retting, R.A. (2004), trafﬁc merging strategy for communication-and sensor- “Types and characteristics of ramp-related motor vehicle enabled vehicles”, 16th International IEEE Conference on crashes on urban interstate roadways in Northern Virginia”, Intelligent Transportation Systems (ITSC 2013). IEEE, Journal of Safety Research, Vol. 35 No. 1, pp. 107-114. pp. 1468-1474. Madhusudhanan, A.K. (2019), “A method to improve an Bellem, H., Schönenberg, T., Krems, J.F. and Schrauf, M. electric vehicle’s range: efﬁcient cruise control”, European (2016), “Objective metrics of comfort: developing a driving Journal of Control, Vol. 48, pp. 83-96. style for highly automated vehicles”, Transportation Research Mohammed, H., Bigazzi, A.Y. and Sayed, T. (2019), Part F: Trafﬁc Psychology and Behaviour, Vol. 41, pp. 45-54. “Characterization of bicycle following and overtaking Bellem, H., Thiel, B., Schrauf, M. and Krems, J.F. (2018), maneuvers on cycling paths”, Transportation Research Part C: “Comfort in automated driving: an analysis of preferences Emerging Technologies, Vol. 98, pp. 139-151. for different automated driving styles and their dependence Nalic, D., Pandurevic, A., Eichberger, A. and Rogic, B. on personality traits”, Transportation Research Part F: trafﬁc (2020b), “Design and implementation of a co-simulation Psychology and Behaviour, Vol. 55, pp. 90-100. framework for testing of automated driving systems”, Bieker-Walz, L., Behrisch, M. Junghans, M. and Gimm, K. Sustainability, Vol. 12 No. 24, p. 10476. (2017), “Evaluation of car-following-models at controlled Nalic, D., Li, H., Eichberger, A., Wellershaus, C., Pandurevic, intersections”. A. and Rogic, B. (2020a), “Stress testing method for 44 Effects of connected and autonomous vehicle merging Journal of Intelligent and Connected Vehicles Lishengsa Yue, Mohamed Abdel-Aty and Zijin Wang Volume 5 · Number 1 · 2022 · 36–45 scenario-based testing of automated driving systems”, IEEE using Gaussian ﬁnite mixture models”, The R Journal, Vol. 8 Access, Vol. 8, pp. 224974-224984. No. 1, p. 289. Nandi, A.K., Chakraborty, D. and Vaz, W. (2015), “Design of Sefati, M., Chandiramani, J., Kreisköther, K., Kampker, A. a comfortable optimal driving strategy for electric vehicles and Baldi, S. (2017), “Towards tactical behaviour planning using multi-objective optimization”, Journal of Power Sources, under uncertainties for automated vehicles in urban Vol. 283, pp. 1-18. scenarios”, 2017 IEEE 20th International Conference on Nguyen, N.A., Moser, D., Schrangl, P., del Re, L. and Jones, Intelligent Transportation Systems (ITSC). IEEE, pp. 1-7. S. (2017), “Autonomous overtaking using stochastic model Treiber, M., Hennecke, A. and Helbing, D. (2000), predictive control”, 2017 11th Asian Control Conference “Congested trafﬁc states in empirical observations and (ASCC). IEEE, pp. 1005-1010. microscopic simulations”, Physical Review E,Vol.62 No. 2, Olofsson, M. and Pettersson, J. (2015), Parameterization and p. 1805. Validation of Road and Driver Behavior Models for Carmaker Wang, Y., Wenjuan, E., Tang, W., Tian, D., Lu, G. and Yu, G. Simulations and Transmission HIL-Rig. (2013), “Automated on-ramp merging control algorithm Potzy, J., Feinauer, S., Siedersberger, K.-H. and Bengler, K. based on internet-connected vehicles”, IET Intelligent (2019), “Manual drivers’ evaluation of automated merging Transport Systems, Vol. 7 No. 4, pp. 371-379. behavior in dense trafﬁc: efﬁciency matters”, 2019 IEEE Weng, J. and Meng, Q. (2014), “Rear-end crash potential Intelligent Transportation Systems Conference (ITSC). IEEE, estimation in the work zone merging areas”, Journal of pp. 3454-3460. Advanced Transportation, Vol. 48 No. 3, pp. 238-249. Reinolsmann, N., Alhajyaseen, W., Brijs, T., Pirdavani, A., Weng, J., Xue, S., Yang, Y., Yan, X. and Qu, X. (2015), “In- Hussain, Q. and Brijs, K. (2019), “Investigating the impact depth analysis of drivers’ merging behavior and rear-end of dynamic merge control strategies on driving behavior on crash risks in work zone merging areas”, Accident Analysis & rural and urban expressways – a driving simulator study”, Prevention, Vol. 77, pp. 51-61. Transportation Research Part F: Trafﬁc Psychology and Wu, P., Gao, F. and Li, K. (2019), “A vehicle type dependent Behaviour, Vol. 65, pp. 469-484. car-following model based on naturalistic driving study”, Samiee, S.,Azadi,S., Kazemi, R.and Eichberger,A. (2016), Electronics, Vol. 8 No. 4, p. 453. “Towards a decision-making algorithm for automatic lane Yue, L., Abdel-Aty, M., Wu, Y., Yuan, J. and Morris, M. change manoeuvre considering trafﬁcdynamics”, PROMET – (2020), “Inﬂuence of pedestrian-to-vehicle technology on Trafﬁc&Transportation,Vol.28No.2,pp.91-103. drivers’ response and safety beneﬁts considering pre-crash Schreiber, V., Ivanov, V., Augsburg, K., Noack, M., Shyrokau, conditions”, Transportation Research Part F: Trafﬁc Psychology B., Sandu, C. and Els, P.S. (2018), “Shared and distributed and Behaviour, Vol. 73, pp. 50-65. X-in-the-loop tests for automotive systems: feasibility study”, IEEE Access, Vol. 6, pp. 4017-4026. Corresponding author Scrucca, L., Fop, M., Murphy, T.B. and Raftery, A.E. (2016), Lishengsa Yue can be contacted at: 2017lishengsa@knights. “Mclust 5: clustering, classiﬁcation and density estimation ucf.edu For instructions on how to order reprints of this article, please visit our website: www.emeraldgrouppublishing.com/licensing/reprints.htm Or contact us for further details: permissions@emeraldinsight.com

Journal

Journal of Intelligent and Connected Vehicles – Emerald Publishing

Published: Feb 17, 2022

Keywords: Driving simulator; Connected and autonomous vehicle; Merging algorithm; Merging behavior; Safety and comfort; Driving safety; Driving comfort

References

Modeling driver behavior and safety on freeway merging areas

Ahammed, M.A.; Hassan, Y.; Sayed, T.A.
Hardware-in-the-loop test of an open-loop fuzzy control method for decoupled electrohydraulic antilock braking system

Aksjonov, A.; Ricciardi, V.; Augsburg, K.; Vodovozov, V.; Petlenkov, E.
Advanced co-simulation framework for cooperative maneuvers among vehicles

Artunedo, A.; Godoy, J.; Haber, R.; Villagrá, J.; del, T.,R.M.
A k-leader fuel-efficient traffic model

Awal, T.; Kulik, L.; Ramamohanrao, K.
Optimal traffic merging strategy for communication-and sensor-enabled vehicles

Awal, T.; Kulik, L.; Ramamohanrao, K.
Objective metrics of comfort: developing a driving style for highly automated vehicles

Bellem, H.; Schönenberg, T.; Krems, J.F.; Schrauf, M.
Comfort in automated driving: an analysis of preferences for different automated driving styles and their dependence on personality traits

Bellem, H.; Thiel, B.; Schrauf, M.; Krems, J.F.
Evaluation of car-following-models at controlled intersections

Bieker-Walz, L.; Behrisch, M.; Gimm, K.
Coordinated merge control based on V2V communication

Chou, F.-C.; Shladover, S.E.; Bansal, G.
Traffic studies at T-Junctions. 2: A conflict simulation record

Cooper, D.F.; Ferguson, N.
Motion planning for autonomous vehicles considering longitudinal and lateral dynamics coupling

Dang, D.; Gao, F.; Hu, Q.
A rule-based cooperative merging strategy for connected and automated vehicles

Ding, J.; Li, L.; Peng, H.; Zhang, Y.
Validation of the microscopic traffic flow model VISSIM in different real-world situations

Fellendorf, M.; Vortisch, P.
Near miss determination through use of a scale of danger

Hayward, J.C.
Generalized force model of traffic dynamics

Helbing, D.; Tilch, B.
Cooperative merging control via trajectory optimization in mixed vehicular traffic

Karimi, M.; Roncoli, C.; Alecsandru, C.; Papageorgiou, M.
End-to-end reinforcement learning for autonomous longitudinal control using advantage actor critic with temporal context

Kuutti, S.; Bowden, R.; Joshi, H.; de Temple, R.; Fallah, S.
Efficient control of fully automated connected vehicles at freeway merge segments

Letter, C.; Elefteriadou, L.
Longitudinal control algorithm for automated vehicle merging

Lu, X.-Y.; Hedrick, J.K.
Automated vehicle merging maneuver implementation for AHS

Lu, X.-Y.; Tan, H.-S.; Shladover, S.E.; Hedrick, J.K.
Types and characteristics of ramp-related motor vehicle crashes on urban interstate roadways in Northern Virginia

McCartt, A.T.; Northrup, V.S.; Retting, R.A.
A method to improve an electric vehicle’s range: efficient cruise control

Madhusudhanan, A.K.
Characterization of bicycle following and overtaking maneuvers on cycling paths

Mohammed, H.; Bigazzi, A.Y.; Sayed, T.
Design and implementation of a co-simulation framework for testing of automated driving systems

Nalic, D.; Pandurevic, A.; Eichberger, A.; Rogic, B.
Stress testing method for scenario-based testing of automated driving systems

Nalic, D.; Li, H.; Eichberger, A.; Wellershaus, C.; Pandurevic, A.; Rogic, B.
Design of a comfortable optimal driving strategy for electric vehicles using multi-objective optimization

Nandi, A.K.; Chakraborty, D.; Vaz, W.
Autonomous overtaking using stochastic model predictive control

Nguyen, N.A.; Moser, D.; Schrangl, P.; del Re, L.; Jones, S.
Manual drivers’ evaluation of automated merging behavior in dense traffic: efficiency matters

Potzy, J.; Feinauer, S.; Siedersberger, K.-H.; Bengler, K.
Investigating the impact of dynamic merge control strategies on driving behavior on rural and urban expressways – a driving simulator study

Reinolsmann, N.; Alhajyaseen, W.; Brijs, T.; Pirdavani, A.; Hussain, Q.; Brijs, K.
Towards a decision-making algorithm for automatic lane change manoeuvre considering traffic dynamics

Samiee, S.; Azadi, S.; Kazemi, R.; Eichberger, A.
Shared and distributed X-in-the-loop tests for automotive systems: feasibility study

Schreiber, V.; Ivanov, V.; Augsburg, K.; Noack, M.; Shyrokau, B.; Sandu, C.; Els, P.S.
Mclust 5: clustering, classification and density estimation using Gaussian finite mixture models

Scrucca, L.; Fop, M.; Murphy, T.B.; Raftery, A.E.
Towards tactical behaviour planning under uncertainties for automated vehicles in urban scenarios

Sefati, M.; Chandiramani, J.; Kreisköther, K.; Kampker, A.; Baldi, S.
Congested traffic states in empirical observations and microscopic simulations

Treiber, M.; Hennecke, A.; Helbing, D.
Automated on-ramp merging control algorithm based on internet-connected vehicles

Wang, Y.; Wenjuan, E.; Tang, W.; Tian, D.; Lu, G.; Yu, G.
Rear‐end crash potential estimation in the work zone merging areas

Weng, J.; Meng, Q.
In-depth analysis of drivers’ merging behavior and rear-end crash risks in work zone merging areas

Weng, J.; Xue, S.; Yang, Y.; Yan, X.; Qu, X.
A vehicle type dependent car-following model based on naturalistic driving study

Wu, P.; Gao, F.; Li, K.
Influence of pedestrian-to-vehicle technology on drivers’ response and safety benefits considering pre-crash conditions

Yue, L.; Abdel-Aty, M.; Wu, Y.; Yuan, J.; Morris, M.

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Effects of connected and autonomous vehicle merging behavior on mainline human-driven vehicle

Effects of connected and autonomous vehicle merging behavior on mainline human-driven vehicle

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Effects of connected and autonomous vehicle merging behavior on mainline human-driven vehicle

Effects of connected and autonomous vehicle merging behavior on mainline human-driven vehicle

Abstract

Journal

Recommended Articles

References

Our policy towards the use of cookies