Strona główna Odkrywaj Pomoc
Zaloguj się
ardo
/
Obsidian
1
0
Forkuj 0
Pliki Problemy 0 Oczekujące zmiany 0 Wiki
Drzewo: a0e0ef701f
Gałęzie Tagi
Obsidian
/
Zotero
/
001_artiklid
/
Road Safety Indicators
/
olufowobisafetyeffectivenessforwardcollision2025.md

olufowobisafetyeffectivenessforwardcollision2025.md 59 KB
Historia Czysty

category: literaturenote citekey: olufowobisafetyeffectivenessforwardcollision2025 title: "Safety effectiveness of forward collision warning systems in the vehicle fleet: A driving simulation study" authors: "Olufowobi, Oluwaseun; Ivan, John; Wang, Kai; Eluru, Naveen" year: 2025 date: 2025-00-00 2025 doi: 10.1016/j.aap.2025.108078 publication: "Accident Analysis & Prevention" url: "https://doi.org/10.1016/j.aap.2025.108078" zotero_key: PU33VT7Q zotero_storage: M4N3UNWT collections: transpordi_planeerimine folder: 001_artiklid/Road Safety Indicators firstAuthor: "Olufowobi, Oluwaseun"

status: converted

Contents lists available at ScienceDirect

Accident Analysis and Prevention

journal homepage: www.elsevier.com/locate/aap

Safety effectiveness of forward collision warning systems in the vehicle fleet: A driving simulation study☆

Oluwaseun Olufowobi a,* , John Ivan a , Kai Wang b , Naveen Eluru c

  • a Department of Civil and Environmental Engineering, University of Connecticut, Storrs CT, USA
  • b Connecticut Transportation Safety Research Center, University of Connecticut, Storrs CT, USA
  • c Department of Civil, Environmental and Construction Engineering, University of Central Florida, Orlando, FL, USA

ARTICLE INFO

Keywords: Rear-end crashes Forward Collision Warning Driving Simulation Traffic Conflict

ABSTRACT

The increasing rate of crashes globally has prompted the development of strategies like the Advanced Driver Assistance systems (ADAS) to improve safety. These systems range from subtle speed adjustment alerts to automatic emergency braking. One such system is the forward collision warning (FCW), which aims to mitigate collisions, particularly rear-end crashes, by providing visual, auditory, or tactile alerts of impending collisions. However, there has not been enough attention given to how drivers might react to FCW without unintentionally distracting them, while also using serious conflicts to quantify safety. This driving simulation study aimed to assess the effectiveness of the FCW by examining serious conflict as a surrogate measure for actual crashes. The goal was to estimate a Crash Modification Factor (CMF) for FCW systems within vehicle fleet, considering varying market penetration rates ranging from 10 percent to 50 percent in increments of 10 percent. Scenarios were created where drivers encountered different road and traffic conditions to evaluate their responses to unexpected events. A total of 133 participants completed a between subject design in which half of them drove the course with the FCW programmed into the simulation, and the other half did not have the FCW. The evaluation of serious conflicts utilized the Swedish Traffic Conflict Technique, employing two key indicators: Time to Accident (TA) and Conflicting Speed. Conflict severity was measured by considering two variables − the lane position and speed of the participants which was used to compute the TA. As expected, drivers in each scenario experienced fewer serious conflicts when assisted by FCW systems compared to the drivers without them. The resulting CMFs from the study can be integrated into crash prediction models to support efforts to keep crash prediction models up to date by accounting for the effects of increasing numbers of vehicles having FCW systems.

1. Introduction

One of the significant socio-economic challenges worldwide is the prevalence of traffic crashes. According to a recent report by the National Highway Traffic Safety Administration (NHTSA), 39,221 Americans lost their lives in car crashes in 2022 (National Highway Traffic Safety Administration, 2023), with rear-end collisions comprising nearly one-third of these incidents. Driver inattention was identified as a leading cause in approximately 91 percent of these rear-end collisions. Factors contributing to driver inattention include distractions, fatigue, and environmental conditions like fog or sun glare, which can hinder a driver's ability to detect obstacles ahead (National Transportation Safety Board, 2002). In response to these challenges, various strategies have been developed to assist drivers in preventing collisions from such factors, ranging from subtle speed adjustment alerts to systems that automatically engage emergency braking.

With the advancement of these safety strategies, vehicles equipped with these safety systems are expected to experience a decrease in total crashes. One such strategy is the forward collision warning (FCW) system, which aims to assist drivers in mitigating or preventing collisions, particularly rear-end crashes, by providing visual, auditory, or tactile alerts of potential collisions (Forkenbrock, 2009). According to a 2002 study by Daimler-Chrysler, approximately 60 percent of rear-end collisions could potentially be avoided if drivers had an additional 0.5 s of

☆ This article is part of a special issue entitled: 'RSS 2024' published in Accident Analysis and Prevention.

* Corresponding author.

E-mail addresses: oluwaseun.olufowobi@uconn.edu (O. Olufowobi), john.ivan@uconn.edu (J. Ivan), kai.wang@uconn.edu (K. Wang), Naveen.Eluru@ucf.edu (N. Eluru).

warning, while giving drivers an extra second of warning could potentially prevent about 90 percent of such collisions (Ankrum, 1992). This places FCW high on the list of solutions that can contribute significantly to reducing the number and the severity of crashes.

The significance of studying FCW systems lies in their potential to enhance traffic safety by mitigating high-risk interactions and preventing crashes before they occur. These systems, by providing timely alerts, can help drivers respond more effectively to imminent hazards, particularly in complex urban and rural environments where the variability of driving conditions poses unique challenges. Using methods such as analyzing traffic conflicts and near-misses as surrogates for crashes provides a more comprehensive understanding of FCW system performance. This approach not only addresses the limitations of traditional crash data analysis but also provides valuable insights into driver behavior and the effectiveness of advanced safety systems, supporting the development of safer roads and more impactful interventions.

This paper describes an effort to quantify the effectiveness of FCW systems in reducing serious traffic conflicts by comparing driver performance in vehicles equipped with FCW systems to those without such systems in a driving simulation experiment. Participants were exposed to road and traffic conditions deliberately designed to assess their responses to unexpected forward obstacles when they get warnings from an FCW system or when there is no system. Unlike previous study such as Lee et al. (Lee et al., 2002); which focused on the value of a FCW system in a scenario where the driver is intentionally distracted as part of the experiment, and the hazard was a stopped lead vehicle. Here we explore the value of the FCW system for drivers who may or may not be distracted and where the hazard could be a vehicle that stops or crosses the travel path suddenly, or a non-motorized user crossing the street unexpectedly. By exploring these broader scenarios, this study seeks to provide a more comprehensive evaluation of FCW systems across diverse driving conditions.

2. Literature review

2.1. Forward collision warning systems

FCW systems, which are part of many vehicles' Collision Avoidance Systems (CAS), assist drivers by keeping their attention on the road ahead, helping them to anticipate and respond to potential hazards more effectively. Several studies (Najim et al., 2007; Rakha et al., 2010; Kusano and Gabler, 2012; Seaman et al., 2022) emphasize the safety advantages of these systems to significantly lower the risks of serious and fatal injuries. For example, Lee et al., (Lee et al., 2002) exposed driving simulator users to a scenario with a stopped lead vehicle where the driver was intentionally distracted and found that the FCW system reduced the number of collisions for that scenario by 80.7 percent. Similarly, Teoh (Teoh, 2021) estimated the effectiveness of FCW using detailed data from exposure measures extracted from video footage and concluded that FCW was associated with a statistically significant 22 percent reduction in the rate of police-reportable crashes per vehicle miles traveled. These findings complement estimates from Jermakian (Jermakian, 2011) that FCW systems could prevent up to 1.2 million crashes annually in the U.S., or 20 percent of all police-reported crashes, and Kusano and Gabler (Kusano and Gabler, 2012) indicating dramatic reductions in serious and fatal injuries during rear-end collisions using data extracted from the National Automotive Sampling System/Crashworthiness Data System (NASS/CDS). All this research highlights the critical role of FCW systems in improving roadway safety.

The success of FCW systems, however, depends significantly on the timing of their warning, a critical aspect in human factors research. Studies such as McLaughlin et al., (McLaughlin et al., 2008) and Abe and Richardson (Abe and Richardson, 2008) noted that providing early warning has an enormous potential to prevent crashes. Moreover, other studies have explored how various driving situations and history of hazard warning can influence drivers' response to warning and

hazardous situations. Failures of the system to produce warnings that drivers find useful (incorrect warning) and understandable may diminish the driver's reliance on and compliance with warnings (Seaman et al., 2022). To address these challenges, researchers like Kiefer et al. (Kiefer et al., 1999) developed FCW alert timing algorithms based on comprehensive human factors research. These algorithms account for the time required for a driver to perceive and react to an alert, as well as the time needed to execute braking or other evasive actions. Consequently, since FCW alerts are designed to provide drivers with sufficient lead time, unlike other CAS, such as Automatic Emergency Braking (AEB), which activate later in the collision sequence when the driver fails to respond (Kiefer et al., 1999). This proactive approach to alert timing highlights the importance of early intervention in mitigating potential crashes.

Despite its strengths, FCW systems face limitations in certain scenarios. For instance, in the NHTSA FCW performance test procedure, drivers follow a moving vehicle that suddenly changes lanes, exposing a stopped vehicle ahead. Here, the FCW system issues an alert at a time-tocollision (TTC) of 2.1 s to prompt the driver, who may be distracted, to respond and prevent or mitigate a collision. While FCW systems succeed in initiating driver responses like braking or evasive maneuvers, Kiefer et al., (Kiefer et al., 1999) noted that AEB systems may not be effective in such scenarios. This limitation arises from factors such as limited target acquisition time, sensor constraints, and delays in algorithm processing and brake activation. These findings highlight the need to optimize FCW timing and reliability to maximize safety outcomes and maintain driver trust. By addressing these limitations and refining system responses, FCW systems can enhance their role in preventing crashes and improving overall road safety.

Research has also explored the role of market penetration rates (MPR) in increasing the safety benefits of advanced technologies like FCW. For example, Xiao et al. (Xiao et al., 2021) conducted a *meta*analysis evaluating crash reduction by MPR. The results indicated that the number of conflicts is exponentially reduced as MPR goes up, safety is enhanced by 4 percent with an MPR of 10 percent, and by 43 percent with 90 percent MPR. Furthermore, Papadoulis et al. (Papadoulis et al., 2019) evaluated the safety impact of connected vehicles (CVs) using traffic simulation. The results showed that CVs bring about compelling benefits to road safety as traffic conflicts significantly reduce even at low MPRs. Specifically, estimated traffic conflicts were reduced by 12–47 percent, 50–80 percent, 82–92 percent and 90–94 percent for 25 percent, 50 percent, 75 percent, and 100 percent CVs penetration rates, respectively.

Together, this body of research illustrates the widespread benefits of FCW systems, from individual crash mitigation to large-scale reductions in crash severity and frequency. Despite the ongoing advancements in other technologies, FCWs continues to play a crucial role in crash avoidance and mitigation. To further evaluate their performance and ability to mitigate collisions, surrogate safety assessment offers a valuable approach by analyzing traffic conflicts, which often precedes crashes..

2.2. Crashes and serious conflicts

The conventional approach to traffic safety has primarily focused on the occurrence of traffic crashes and their outcomes (Lai et al., 2014). In contrast, surrogate safety assessment offers an alternative method by focusing on the analysis of safety–critical events such as traffic conflicts to evaluate safety. The concept of traffic conflicts was first introduced by (Klebelsberg, 1964) as cited by Tarko (Tarko, 2018), who defined them as dangerous traffic interactions. Laureshyn and Varhelyi (Laureshyn and Varhelyi, 2018) defined a traffic conflict as "an observable situation in which two or more road users approach each other in space and time to such an extent that there is a risk of collision if their movements remain unchanged". The proximity of road users to each other could be physically measured in temporal and/or spatial dimensions, and

thresholds could be used to identify conflicts. Perkins and Harris (Perkins and Harris, 1968) expanded on the concept defining traffic conflicts as any situations necessitating evasive actions such as braking. According to this definition, it implies that conflicts and crashes are of a similar nature except for the presence and the success of an evasive action (Lai et al., 2014). Traffic conflicts offer significant advantages for road safety analysis compared to collisions. They occur more frequently, are easier to observe, and involve lower costs (Ismail et al., 2011).

Various conflict indicators have been developed to measure the severity of an interaction by quantifying the spatial and temporal proximity of two or more road users (Arun et al., 2021). These indicators are classified into two categories: Spatial/temporal proximities and evasive action indicators. Commonly used measures of conflict severity include conflicting speed, TTC, Post-Encroachment Time (PET), and the severity index (Lai et al., 2014; Tarko, 2018). Brown (Brown, 1994) and Tarko et al., (Tarko, 2021) provide comprehensive summaries of these indicators. The TTC indicator, used in this study, is defined as "…the time that remains until a collision between two vehicles would have occurred if the collision course and speed difference are maintained" (Tarko, 2021). The main advantage of conflict indicators is their ability to capture the severity of an interaction in an objective and quantitative way. Numerous studies have used traffic conflicts together with crashes for safety analysis and have shown that conflicts can adequately serve as a proxy for crashes (Lai et al., 2014; Tarko et al., 2009; Autey et al., 2012; Essa and Sayed, 2020; Charly and Mathew, 2019).

Although previous studies have contributed to examining the effectiveness of the FCW, few studies have explored estimating serious conflict with the aim of providing information for crash modification factors (CMF) for the FCW systems. Near crashes, also known as serious conflicts, will be used as a surrogate for crashes to represent quantitative safety, because crashes are expected to be extremely rare given the number of drivers that will be observed in the experiment and primarily because accurately identifying crashes which are considered the gold standard epidemiological measure cannot be identified within the constraint of a driving simulator.

3. Materials and method

3.1. Participants

In total, 142 individuals took part in the study, with 64 females, 77 males, and one person who chose not to disclose their gender. Nine participants were unable to complete the experiment due to simulator sickness. Of the 133 who finished, 58 (41 percent) were between 18–29 years old, 55 (41 percent) were between 30–64, and 20 (15 percent) were 65 or older. A power analysis based on previous research (Gaspar et al., 2015) suggested that 36 pairs of participants (72 total) would be needed to achieve adequate statistical power (β = 0.8). However, considering potential dropouts due to simulator sickness, the estimate increased to 57 pairs (114 participants) if the dropout risk was 36 percent, and up to 71 pairs (142 participants) if the dropout rate reached 51 percent.

To ensure a balanced demographic, a representative sample of demographic were recruited which were deliberately distributed among the scenarios. All participants held valid driver's licenses and were in good physical health. Recruitment was carried out by posting flyers on University of Connecticut (UConn) bulletin boards, adhering to Institutional Review Board (IRB) standards, and distributing flyers at senior and community centers to attract older drivers and residents not affiliated with the university.

3.2. Equipment

The study utilized the Realtime Technologies RDS-2000 Full Cab with Segmented Screen Driving Simulator, along with multiple cameras and four projectors (rear, front, right side, and left side projections) to create a highly realistic virtual environment. This simulator is housed and maintained by the Connecticut Transportation Safety Research Center at UConn. Some of the features of the driving simulator include the Internet Scene Assembler®, used for modification of the virtual environment; and SimObserver®, integrated with the virtual environment and used for data and video synchronization, video capture and after-action review. Fig. 1 shows the picture of the UConn driving simulator. The first computer on the left side was used to run the simulator and edit the model, while the computer on the right-hand side was used to extract the data and video clips of the main driving experiment.

3.3. Procedure and experimental design

Upon arrival, participants were provided with an informed consent form to read and sign, which also included their consent for video recording by initialing the document. Following this, participants filled out a questionnaire containing demographic inquiries such as gender, age, race/ethnicity, and years of driving experience. Subsequently, each participant underwent a driving simulator training session lasting approximately five (5) minutes to familiarize themselves with the simulator. An additional five-minute training session was available upon request.

After the training session, half of the participants drove the experiment with an FCW system integrated into the simulator, which provided them with both an audible warning and a visual indicator (on the dashboard) about potential collisions. The remaining participants did not have the FCW system and received no warnings during their drive. Prior to commencing the driving session, participants using the FCW system were informed that their vehicle would produce an audible sound and/or display a flashing indicator on the dashboard to alert them of impending forward collisions. Each participant then engaged in one of four predefined driving scenarios encompassing either urban or rural driving, with the scenario lasting approximately 15 min. Participants were instructed to drive the course as they would in a real-life situation and were told that the purpose of the experiment is to observe their reactions to the scenario they drive.

During the experiment, participants were asked to follow the designated route as a recorded message that was broadcast through speakers in the car provided directions. At some point during the scenario, a stimulus was introduced that necessitated a sudden, unexpected braking maneuver by the driver. For instance, this could involve a parked vehicle suddenly pulling in front without warning, or a vehicle, animal, or pedestrian crossing the vehicle's path.

Upon completing their turns in the driving simulator, participants were debriefed by a researcher and given a debriefing form. Furthermore, the researcher verbally explained the study and its objectives. Since the experiment did not involve any deception, this debriefing was purely informational.

3.4. Driving scenarios

Driving scenarios were created for both urban and rural environments to reflect real-world conditions. The urban environment (Fig. 2) consisted of an undivided roadway section with two lanes of traffic in both directions. The roadway included all the typical features of an urban road, such as speed limit signs, traffic signals, parked cars/ trucks on the shoulders, and had paved curbs with pedestrian movements on both sides of the road. The rural scenario (Fig. 3) consisted of a two-lane undivided roadway section with traffic simulated into the experiment to make it feel like real-world driving. The rural setting was a mix of countryside and residential areas with the road primarily straight, with a few horizontal curves.

Table 1 illustrates the demographic distribution of participants across each scenario. Participants were assigned based on their demographic information, ensuring a balanced distribution of age, gender,

Fig. 1. UConn Driving Simulator.

Fig. 2. Photo. Section of urban roadway used in experiment.

Fig. 3. Photo. Section of rural roadway used in experiment.

and years of driving experience in each group with and without the FCW system. Efforts were also made to maintain a balanced distribution of race/ethnicity in each group as practical. The study employed a between-subjects design with the driving simulator equipped with an FCW system as mentioned earlier, which operates by continuously monitoring the road ahead using a combination of sensors and cameras. When a potential collision is detected, the system calculates the TTC based on the vehicle's speed and distance relative to the object ahead. If the TTC falls below a certain threshold, the system issues a warning to the driver. These warnings are intended to give the driver sufficient time to react and take action to avoid a collision. Both visual and auditory alerts were triggered when the distance became too short to avoid a crash. The FCW visual display, located on the dashboard in front of the driver, showed a flashing red icon of two cars in proximity when activated, while a flashing yellow pedestrian icon appeared for potential pedestrian collisions. This design was chosen because most FCW systems on the market utilize a combination of visual and auditory alerts (Forkenbrock, 2009), with the visual display typically located in the instrument cluster near the speedometer.

3.5. Data collection

To evaluate conflict severity, two primary variables were required: time to accident (TA) and conflicting speed (CS). These variables are further detailed in the data analysis section. TA was calculated using the vehicle's speed and the distance to the collision point, with speed directly obtained from the driving simulator as the participant's driving velocity. Additionally, lane position, measured in meters from the center of the road, was used to represent lateral control. Data were recorded at a frequency of 60 Hz using the simulator's proprietary SimObserver software.

Lane position data were captured for each participant in both "with system" and "without system" scenarios, providing the lateral position at the moment the experimental event occurred. This information was essential for determining the distance to the collision point. Speed data represented the participant's velocity immediately before performing evasive actions, such as braking or swerving.

Evasive actions were identified using video footage from SimObserver, cross-referenced with steering and brake inputs captured by the Data Distillery software. A reduction in steering angle or brake application values was used to pinpoint when participants initiated evasive maneuvers.

The distance to a potential collision was calculated as the gap, in meters, between the vehicle's position during the evasive action and the location of the event along the road. This calculation was used to estimate conflict severity. While this study did not focus on the specific avoidance strategies employed (e.g., braking versus swerving), the actions provided critical data for determining the proximity to potential collisions. Fig. 4 shows a graphical description of how these variables were collected. This systematic data collection ensured accurate measurement of critical variables, enabling a robust evaluation of conflict severity under both experimental conditions (Fig. 5).

Table 1 Driving simulator participant statistics by scenario.

Gender Age Group Scenario 1 Scenario 2 Scenario 3 Scenario 4 Total
With Without With Without With Without With Without
Male 18–29 4 4 4 4 2 2 4 4 28
30–64 4 5 4 3 2 3 3 3 27
65+ 1 1 0 1 3 2 1 0 9
Female 18–29 2 3 4 3 4 5 4 4 29
30–64 3 1 5 5 4 3 2 3 26
65+ 1 1 2 3 2 2 1 1 13
Total 15 15 19 19 17 17 15 15 132

Fig. 4. Graphical Representation of Scenarios and Events.

Fig. 5. Variables of interest from the simulator.

4. Data analysis

To analyze the FCW systems in mitigating traffic conflicts, data collected from the driving simulator were processed and evaluated using a traffic conflict methodology. Additionally, chi-square analysis was employed to compare serious conflict frequencies between scenarios with and without FCW systems. The results of the chi-square test validate the findings, ensuring their robustness and providing justification for calculating CMFs (Fig. 6).

4.1. Statistical analysis

To statistically evaluate the differences in serious conflict frequencies between the with FCW and without FCW scenarios, a chisquare test of independence was performed prior to calculating the CMF to validate the findings and ensure they justified the CMFs calculations. The analysis was conducted at two levels:

Fig. 6. Conflict severity diagram (Laureshyn, A., & Varhelyi, A., 2018).

Aggregated Analysis: Combined data from all scenarios to assess the overall impact of the FCW system.

Scenario-Specific Analysis: Evaluated whether the effectiveness of the FCW system varied across the different simulated scenarios.

The steps in the chi-square analysis were as follows:

  • Contingency Table Construction: A table was created to record the number of traffic conflicts in each category (e.g., "with FCW" and "without FCW").
  • Chi-Square Statistic Calculation: The chi-square statistic (χ2) was calculated using equation (1):

$$\chi 2 = \sum \frac{(O-E)^2}{E} \tag{1}$$

• Significance Testing: The calculated χ2 value were compared against the critical value from the chi-square distribution table at a significance level of 0.05 to determine whether differences were statistically significant.

This two-level analysis enabled a comprehensive evaluation of the FCW system's effectiveness, both overall and within specific scenarios, providing insight into its impact on traffic conflict severities.

4.2. Traffic conflict methodology

This study utilized the Swedish Traffic Conflict Technique (TCT) to evaluate traffic conflicts using two indicators: the time to accident (TA) value and the conflicting speed (CS). The TA is the time between when a particular road user successfully performs an evasive action and when the collision would have occurred if they had continued with unchanged speeds and direction, while the CS is the speed of the road user at the moment just before taking the evasive action (Laureshyn and Varhelyi, 2018). The TA and CS values are used to define the conflict severity. The conflicting speed affects the outcome of the collision (the resulting pedestrian injury depends on the speed), and a higher speed would require longer stopping time and distance. Thus, a higher CS would indicate a more severe conflict for the same TA (Laureshyn and Varhelyi, 2018). Because it can be challenging to directly estimate the TA in live traffic, it can be estimated from the vehicle speed and the distance to the collision point using Table 2.

The conflict severity is then determined from the TA and the CS according to the graph in Fig. 2.

Conflicts with a severity level above 26 are ranked as serious. For instance, a vehicle approaching 30 mph (48.3 km/hr.) with a TA of 2.8 sec would pass the potential conflict point with too much time remaining for the interaction to be classified as a serious conflict. However, a 45 mph (72.4 km/hr.) approach speed would just exceed that level. These severity levels can be used to understand serious conflicts to estimate the CMF for FCW systems. A CMF is the ratio of the number of crashes expected with a countermeasure to the number of crashes expected without it. In other words, when implementing a countermeasure, multiplying the existing expected number of crashes by the CMF gives an estimate of the expected number of crashes after implementing the countermeasure. CMFs with a value less than 1.0 indicate an expected decrease while a value greater than 1.0 indicates an expected increase in crashes. The FCW system can be said to be a countermeasure installed in vehicles to help mitigate the occurrence of crashes in vehicles that have it installed. A CMF for this "countermeasure" would have to also account for the penetration rate of the FCW system in the vehicle fleet. Estimating a CMF value for this FCW system would be helpful for adjusting the expected number of crashes in future conditions as the proportion of vehicles with such systems increases.

Following are the steps to estimate the CMF:

● Estimate a ratio to account for the safety effect of having a FCW system (RFCW),

$$R{FCW} = \frac{n{FCW}}{n_o} \tag{2}$$

where nFCW is the number of serious conflicts in vehicles with the FCW and no is the number of serious conflicts in vehicles without the FCW.

  • Define the proportion of the vehicle fleet with the FCW system (PFCW). Here, five different proportions of the vehicle fleet from 10 to 50 percent in increments of 10 percent were investigated.
  • Calculate the estimated CMF for the FCW system:

$$CMF{FCW} = 1 + P{FCW}(R_{FCW} - 1)$$ (3)

Table 2 TA values estimated from vehicle speed and distance to collision point (Laureshyn, A., & Varhelyi, A., 2018).

Speed Distance, m
Km
/h		 0.5 1 2 3 4 5 6 7 8 9 10 15 20 25 30 35 40 45 50 55
5 0.4 0.7 1.4 2.2 2.9 3.6 4.3 5.0 5.8 6.5 7.2
10 0.2 0.4 0.7 1.1 1.4 1.8 2.2 2.5 2.9 3.2 3.6 5.4 7.2 9.0
15 0.1 0.2 0.5 0.7 1.0 1.2 1.4 1.7 1.9 2.2 2.4 3.6 4.8 6.0 7.2 8.4 9.6
20 0.1 0.2 0.4 0.5 0.7 0.9 1.1 1.3 1.4 1.6 1.8 2.7 3.6 4.5 5.4 6.3 7.2 8.1 9.0 9.9
25 0.1 0.1 0.3 0.4 0.6 0.7 0.9 1.0 1.2 1.3 1.4 2.2 2.9 3.6 4.3 5.0 5.8 6.5 7.2 7.9
30 0.1 0.1 0.2 0.4 0.5 0.6 0.7 0.8 1.0 1.1 1.2 1.8 2.4 3.0 3.6 4.2 4.8 5.4 6.0 6.6
35 0.1 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.5 2.1 2.6 3.1 3.6 4.1 4.6 5.1 5.7
40 0.0 0.1 0.2 0.3 0.4 0.5 0.5 0.6 0.7 0.8 0.9 1.4 1.8 2.3 2.7 3.2 3.6 4.1 4.5 5.0
45 0.1 0.2 0.2 0.3 0.4 0.5 0.6 0.6 0.7 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0 4.4
50 0.1 0.1 0.2 0.3 0.4 0.4 0.5 0.6 0.6 0.7 1.1 1.4 1.8 2.2 2.5 2.9 3.2 3.6 4.0
55 0.1 0.1 0.2 0.3 0.3 0.4 0.5 0.5 0.6 0.7 1.0 1.3 1.6 2.0 2.3 2.6 2.9 3.3 3.6
60 0.1 0.1 0.2 0.2 0.3 0.4 0.4 0.5 0.5 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7 3.0 3.3
65 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 0.5 0.6 0.8 1.1 1.4 1.7 1.9 2.2 2.5 2.8 3.0
70 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 0.5 0.5 0.8 1.0 1.3 1.5 1.8 2.1 2.3 2.6 2.8
75 0.0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 0.5 0.7 1.0 1.2 1.4 1.7 1.9 2.2 2.4 2.6
80 0.0 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 0.5 0.7 0.9 1.1 1.4 1.6 1.8 2.0 2.3 2.5
85 0.0 0.1 0.1 0.2 0.2 0.3 0.3 0.3 0.4 0.4 0.6 0.8 1.1 1.3 1.5 1.7 1.9 2.1 2.3
90 0.0 0.1 0.1 0.2 0.2 0.2 0.3 0.3 0.4 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
95 0.0 0.1 0.1 0.2 0.2 0.2 0.3 0.3 0.3 0.4 0.6 0.8 0.9 1.1 1.3 1.5 1.7 1.9 2.1
100 0.0 0.1 0.1 0.1 0.2 0.2 0.3 0.3 0.3 0.4 0.5 0.7 0.9 1.1 1.3 1.4 1.6 1.8 2.0

5. Results

5.1. Chi square analysis result

5.1.1. Aggregated analysis

The data were aggregated across all scenarios to assess the overall effect of the FCW system. The contingency table summarizing the aggregated data is shown in Table 5.

The Chi-Square statistic (X-squared) of 8.7576 with a degree of freedom (df) of 1 resulted in a p-value of 0.0030. Since the p-value is less than the significance level of 0.05, indicating that there is a statistically significant difference between the "With FCW" and "No FCW" conditions. This result suggests that the FCW system significantly reduces the occurrence of serious conflicts when considering all scenarios combined.

5.1.2. Scenario-Specific analysis

The analysis was extended to determine whether the FCW system's effectiveness varied across individual scenarios. The contingency in Table 6 below summarizes the serious conflicts by scenario.

The Chi-Square statistic (X-squared) of 0.5516 with a degree of freedom (df) of 3 resulted in a p-value of 0.9074. Since the p-value exceed the threshold of 0.05, indicating that there is no statistically significant difference in the number of serious conflicts between the "With FCW" and "No FCW" conditions when analyzed on a scenario-byscenario basis. The findings indicate that the FCW system's effectiveness is consistent across various scenarios, showing slight variation with specific conditions. This suggests that FCW systems are broadly effective and suitable for diverse driving conditions. However, this result could be influenced by factors such as sample size.

5.2. Safety effectiveness of FCW

To calculate the safety effect of having the FCW was determined by calculating the ratio of serious conflict (nFCW) in the experiment with FCW to the ratio of serious conflict without FCW as discussed earlier. Table 3 shows the raw factor for the different scenarios conducted in the experiment. The raw factor was also calculated for all cases with and without FCW to see if there are substantial differences between individual scenarios and overall cases.

The CMF was calculated based on the various penetration rates of the system and are shown in Table 4.

6. Discussion

The objective of this experiment was to examine the effect of the FCW on drivers with the aim of providing data for CMF related to FCW system and to statistically assess the significance of these effects using chi-square analysis. The chi-square analysis revealed significant differences in the occurrence of serious conflicts showing a significant overall reduction in serious conflicts when FCW systems were used (χ2 = 8.76, p < 0.05). This demonstrates that FCW systems consistently contribute to conflict reduction, especially as their penetration rate increases. The scenario-specific chi-square analysis, however, revealed no statistically significant differences in the number of serious conflicts between the "With FCW" and "No FCW" conditions across the individual scenarios (χ2 = 0.55, p > 0.05). Since the p-value exceeds 0.05, this indicates that

Table 3 Aggregated Chi-Square Result.

Condition Serious Conflicts No Conflicts Total Participants
With FCW 24 42 66
No FCW 42 26 68
X-squared (χ2) 8.7576
Degree of freedom 1
p-value 0.0030

Table 4 Scenario specific Chi-Square Result.

| Scenario | Serious Conflicts (With
FCW) | Serious Conflicts (No
FCW) | | | |-------------------------------|---------------------------------|-------------------------------|--|--| | 1: Urban with entering
car | 6 | 9 | | | | 2: Urban with
pedestrian | 6 | 13 | | | | 3: Rural with animal | 7 | 10 | | | | 4: Rural with pedestrian | 5 | 10 | | | | X-squared (χ2) | 0.5516 | | | | | Degree of freedom | 3 | | | | | p-value | 0.9074 | | | |

Table 5 FCW Ratios by Scenario.

Scenario nFCW nO RFCW = nFCW
no
All cases 24 42 0.57
1: Urban with entering car 6 9 0.67
2: Urban with pedestrian 6 13 0.46
3: Rural with animal 7 10 0.70
4: Rural with pedestrian 5 10 0.50

Table 6 CMF Values by FCW Fleet penetration rate.

Scenario PFCW ¼
0.1		 PFCW ¼
0.2		 PFCW ¼
0.3		 PFCW ¼
0.4		 PFCW ¼
0.5
All cases 0.95 0.91 0.87 0.82 0.78
1: Urban with
entering car		 0.96 0.93 0.89 0.86 0.83
2: Urban with
pedestrian		 0.94 0.89 0.83 0.78 0.73
3: Rural with
animal		 0.97 0.94 0.91 0.88 0.85
4: Rural with
pedestrian		 0.95 0.90 0.85 0.80 0.75

the effectiveness of the FCW system does not vary significantly between scenarios, suggesting consistent performance across different conditions regardless of environmental or contextual differences, such as traffic volume or road layout. While the scenario-specific results do not show statistically significant variation, the CMF calculations provided important insights into crash reductions for individual scenarios.

The resulting CMF was 0.95 which translates to a 5 percent reduction in crashes for all the cases scenario conducted in the experiment when the penetration rate is at 10 percent and increases to 22 percent when the penetration rate rises to 50 percent. For Scenario 1, when the FCW penetration rate is 10 percent, we expect a 4 percent reduction in crashes and a 17 percent reduction at a 50 percent penetration rate. In the case of scenario 2, a CMF 0.94 for a 10 percent penetration rate means we expect a 6 percent reduction in crashes with a 27 percent reduction when the penetration rate of FCW is at 50 percent. While in scenario 3, the event with an animal suddenly crossing into the path of the participant, had a CMF of 0.97 which translates to a 3 percent reduction in crashes when the penetration rate is at 10 percent and 0.85, a 15 percent reduction when the penetration rate is at 50 percent finally in the case of rural with pedestrian had a CMF of 0.95, 5 percent reduction in crashes when the penetration rate is at 10 percent and a CMF of 0.75, 25 percent reduction in crashes when the penetration is at 50 percent. Overall, we can notice that as the penetration rate increases, there is a reduction in crashes. This aligns with some previous studies (Najim et al., 2007; Teoh, 2021; Abdulsattar et al., 2018) who also reported a similar conclusion of decreased crashes as the penetration rate of these systems increases.

As expected, drivers in each scenario experienced fewer serious conflicts when assisted by FCW systems compared to the drivers without them, revealing a significant improvement in safety (potential conflict reduction). These warnings benefited drivers by providing timely alerts, enabling quicker reactions to potential collisions, regardless of whether the drivers are distracted or not. Conflicts are reduced more in urban scenarios than in rural ones. This could suggest that the FCW has its highest effectiveness on urban roads compared to rural roads, due to typical aspects of the urban operating environment, such as a greater variety of obstacles (e.g. parked cars and delivery vehicles), higher volume of traffic, frequent traffic signals, frequent stop and go and pedestrian movement, which maximizes the potential for benefits from FCW. Existing literature also supports this finding; for instance, Yue et al., (Yue et al., 2020) in a meta-analysis estimated the practical effectiveness of six ADAS in avoiding crashes across different roadway facilities and reported that FCW had less practical effectiveness in rural roads compared to urban ones. The study also found that the FCW system was more effective for pedestrian-related events, with drivers experiencing fewer conflicts in both urban and rural environment compared to vehicle and animal-related events. This seems intuitive, as drivers are more proactive in avoiding collisions with pedestrians.

7. Conclusions

This study examined the effectiveness of FCW with the aim of providing information for CMF for the FCW systems for five different proportions of the vehicle fleet from 10 to 50 percent in increments of ten percent. Four scenarios were modeled in urban and rural settings where participants were exposed to conditions deliberately designed to evaluate their responses to unexpected events when they get warnings from an FCW system or when there is no system. To achieve this, two variables, the lane position and speed, were used to measure conflict severity, which was used as a proxy for crashes. The lane position represents the position of the vehicle measured from the center of the road. The conflict severity was then determined from the time to collision and the conflicting speed of the road users. A total of 133 participants completed the experiment.

The Chi-Square analysis revealed a significant overall reduction in serious conflicts when FCW systems were used (χ2 = 8.76, p < 0.05). This demonstrates the system's consistent contribution to conflict reduction, particularly as its penetration rate increases. However, scenario-specific Chi-Square results revealed no statistically significant differences between the "With FCW" and "No FCW" conditions in individual scenarios, suggesting consistent performance across varied conditions. The results suggest that the FCW has the potential of reducing the number of crashes in the vehicles that have them installed. The overall CMF for all the cases of scenarios conducted in the experiment was from 0.78 to 0.95 which is about 10 percent to 22 percent reduction in crashes with a R value of 0.67, while the CMF for scenarios 1 and 2 were from 0.83 to 0.96 and 0.73 to 0.94 which is about 4 percent to 17 percent and 6 percent to 27 percent reduction in crashes respectively and had a R value of 0.67 and 0.46. In the rural scenario, the CMF for scenarios 3 and 4 were from 0.85 to 0.97 and 0.75 to 0.95 with an R value of 0.70 and 0.50, respectively. This indicates that the number of crashes reduces by 3–15 percent for scenario 3 and 5–10 percent for scenario 4. The study also reveals that the FCW enhances drivers' response over a range of speeds, strongly suggesting that the FCW can enhance drivers' response in scenarios that could lead to collisions. Conflicts reduced more in urban scenarios than in rural ones suggesting high effectiveness on the urban roads, in addition the FCW system was more effective for pedestrian-related events, with drivers experiencing fewer conflicts in both urban and rural environment compared to vehicle and animal-related events.

This study provides a valuable contribution by estimating CMFs for FCW systems using realistic driving scenarios that substitute serious traffic conflicts for actual crashes. This approach offers an alternative method to assess safety and evaluate the effectiveness of FCW systems allowing for possibility of the resulting CMFs to be incorporated into crash prediction models, to keep these models up to date by accounting for the effects of increasing numbers of vehicles having FCW systems. The study also provides a more holistic assessment of FCW performance by examining a wide range of real-world scenarios including diverse and unexpected road events in both urban and rural settings without deliberately distracting drivers. This broad approach offers a wider understanding of the system's effectiveness across different contexts.

While the findings highlight the substantial benefits of FCW systems, there are still some unanswered questions. Specifically, although the warning enhanced driver's awareness of potential conflict in the simulator, warnings might be perceived as a nuisance in an actual driving situation when they are too sensitive or too early. Further data collection might be required to assess drivers' response to nuisance alarms generated by early warning or if dynamic adjustment of the warning is feasible. Also, a follow up and longer-term study might be helpful to understand the evolution of driver's attitude and to validate the benefits observed in the study with the possible use of naturalistic studies which can provide more insight into the driver's behavior during their everyday trips. Data from these will help validate the benefits of the FCW and assess drivers' attitude to the system compared to the one time the participants are exposed to the FCW system.

CRediT authorship contribution statement

Oluwaseun Olufowobi: Data curation, Methodology, Writing – original draft, Writing – review & editing. John Ivan: Conceptualization, Supervision, Writing – review & editing. Kai Wang: Visualization. Naveen Eluru: Visualization, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The authors would like to gratefully acknowledge FHWA's sponsorship in this research under project number 693JJ31950017. The ideas expressed in this work do not necessarily indicate acceptance by FHWA of the findings and conclusions expressed within.

Data availability

The authors do not have permission to share data.

References

Abdulsattar, H., Mostafizi, A., Wang, H., 2018. Surrogate safety assessment of work zone rear-end collisions in a connected vehicle environment: agent-based modeling framework. J. Transport. Eng. 144 (X). https://doi.org/10.1061/JTEPBS.0000164. Article 0000164.

Abe, G., Richardson, J., 2008. Alarm timing, trust, and driver expectation for forward collision warning systems. Applied Ergonomics 37 (5), 577–586. https://doi.org/ 10.1016/j.apergo.2005.11.001.

Ankrum, D.R., 1992. Smart vehicles, smart roads. Traffic Safety 92 (3), 6–9.

Arun, A., Haque, S.M.M., Bhaskar, A., Washington, S., Sayed, T., 2021. A systematic mapping review of surrogate safety assessment using traffic conflict techniques. Accid. Anal. Prevent. 153, 106016. https://doi.org/10.1016/j.aap.2021.106016.

Autey, J., Sayed, T., Zaki, M.H., 2012. Safety evaluation of right-turn smart channels using automated traffic conflict analysis. Accid. Anal. Prevent. 45, 120–130. https:// doi.org/10.1016/j.aap.2011.11.015.

Brown, G.R., 1994. Traffic conflicts for road user safety studies. Retrieved from Canad. J. Civil Eng. 21 (1), 1–15. http://www.nrcresearchpress.com/loi/cjce.

Charly, A., Mathew, T.V., 2019. Estimation of traffic conflicts using precise lateral position and width of vehicles for safety assessment. Accid. Anal. Prevent. 132, 105264. https://doi.org/10.1016/j.aap.2019.105264.

Essa, M., Sayed, T., 2020. Comparison between surrogate safety assessment model and real-time safety models in predicting field-measured conflicts at signalized