Modeling Driving Performance Using In-Vehicle Speech Data From a Naturalistic Driving Study.

OBJECTIVE: We aimed to (a) describe the development and application of an automated approach for processing in-vehicle speech data from a naturalistic driving study (NDS), (b) examine the influence of child passenger presence on driving performance, and (c) model this relationship using in-vehicle speech data. BACKGROUND: Parent drivers frequently engage in child-related secondary behaviors, but the impact on driving performance is unknown. Applying automated speech-processing techniques to NDS audio data would facilitate the analysis of in-vehicle driver-child interactions and their influence on driving performance. METHOD: Speech activity detection and speaker diarization algorithms were applied to audio data from a Melbourne-based NDS involving 42 families. Multilevel models were developed to evaluate the effect of speech activity and the presence of child passengers on driving performance. RESULTS: Speech activity was significantly associated with velocity and steering angle variability. Child passenger presence alone was not associated with changes in driving performance. However, speech activity in the presence of two child passengers was associated with the most variability in driving performance. CONCLUSION: The effects of in-vehicle speech on driving performance in the presence of child passengers appear to be heterogeneous, and multiple factors may need to be considered in evaluating their impact. This goal can potentially be achieved within large-scale NDS through the automated processing of observational data, including speech. APPLICATION: Speech-processing algorithms enable new perspectives on driving performance to be gained from existing NDS data, and variables that were once labor-intensive to process can be readily utilized in future research.

Seasonal variation in North American level crossing crash rates is due to weather, not day length.

OBJECTIVE: Collisions between road vehicles and trains at level (grade) crossings can be devastating. Injury and economic considerations make prevention efforts of significant interest to society at all levels, and raise important safety concerns. Improving our understanding of the nature and pattern of crashes at level crossings can help inform a variety of types of safety mitigation strategies, including public education, crossing equipment and vehicle design efforts. METHOD: To this end, a database search of Canadian level crossing crashes for the 11-year period between 2007 and 2017 was conducted to confirm a previously identified seasonal variation in the frequency of level crossing crashes. To determine whether the observed winter increase in crashes was due primarily to winter reductions in light levels/day length or to other seasonal weather factors, a subsequent comparison of Canadian data to American Federal Railroad Administration (FRA) crash data was carried out. A separate inferential log-linear model analysis, using season, time of day and crossing protection type, was also used to explore the increase. RESULTS: As expected, the average rate of crossing collisions in Canada increased during winter months compared to non-winter months. While the seasonal pattern was evident in those U.S. states that experience significant changes in weather patterns (i.e., northern states), it was almost completely absent in those states that do not (i.e., southern states). The log-linear model analysis confirmed these findings. CONCLUSIONS: The seasonal variation in North American level crossing crash rate is a result of winter weather conditions, rather than differences in light levels. Future research is planned that will elucidate the specific environmental and human factors contributing to the increase.

Applying principles of fatigue science to accident investigation: Transportation Safety Board of Canada (TSB) fatigue investigation methodology.

Fatigue that is related to the amount and quality of sleep obtained can impair human performance in ways that can lead to accidents. As many transportation industries operate around the clock, fatigue and its effects cannot be eliminated completely; instead, they must be managed. A first step is to document the prevalence and role of fatigue in accidents that occur. The Transportation Safety Board of Canada (TSB) routinely investigates such transportation industry incidents to determine if fatigue was present, if it played a role, and if there were practices in place to effectively manage it and associated risks. Herein, we summarize and describe the TSB's fatigue investigation methodology in the hopes that investigators of other organizations and domains will find the concepts applicable to their operational context.

How shift scheduling practices contribute to fatigue amongst freight rail operating employees: Findings from Canadian accident investigations.

Canada's freight rail system moves 70% of the country's surface goods and almost half of all exports (RAC, 2016). These include dangerous goods. Anonymous survey of freight rail operating employees conducted by the Teamsters Canada Rail Conference (TCRC, 2014) revealed that many do not report getting enough sleep because of their work schedules, and that fatigue may be affecting their performance at work. Besides general impairments in attention and cognitive functioning, fatigue in railway operating employees slows reaction time to safety alarms and impairs conformance to train operating requirements. Shift scheduling practices can contribute to sleep-related fatigue by restricting sleep opportunities, requiring extended periods of wakefulness and by disrupting daily (circadian) rhythms. The primary goal of accident investigation is to identify causal and contributing factors so that similar occurrences can be prevented. A database search of Transportation Safety Board (TSB) rail investigation reports published in the 21-year period from 1995 to 2015 identified 18 that cited sleep-related fatigue of freight rail operating employees as a causal, contributing, or risk finding. This number represents about 20% of TSB rail investigations from the same period in which a human factors aspect of freight train activities was a primary cause. Exploration of accident themes suggests that management of fatigue and shift scheduling in the freight rail industry is a complex issue that is often not conducive to employee circadian rhythms and sleep requirements. It also suggests that current shift scheduling and fatigue management practices may be insufficient to mitigate the associated safety risk. Railway fatigue management systems that are based on the principles of modern sleep science are needed to improve scheduling practices and mitigate the ongoing safety risk.

The common characteristics and behaviors of child occupants in motor vehicle travel.

Objective: Child occupant behavior and head position when travelling in child restraint systems (CRS) may have an effect on injury risk in the event of a motor vehicle crash. The current study aimed to describe the common characteristics and behaviors of child occupants during everyday, real-world motor vehicle travel in a sample of Australian families to identify potential safety implications of observed behaviors and head position within the CRS. Methods: Two instrumented study vehicles were used by 42 families for approximately two weeks. Continuous video and audio data were collected across 1,651 trips (over 600 hours). An online survey provided additional parent, familial and child occupant data. The characteristics and behaviors of 72 child occupants (aged 14 months to 9 years) who travelled in a forward-facing CRS (FFCRS) or a belt-positioning booster seat (BS) were observed and recorded by manual review of a sample of the video/audio recordings. One quarter of all trips (n = 414) was randomly selected for coding/analysis and, within each trip, one child occupant was selected who was travelling in a FFCRS or BS. Child occupant behaviors, head position within the FFCRS or BS, and other relevant information was coded for each trip during nine discrete five second intervals or 'epochs' (5%, 17%, 25%, 30%, 50%, 53%, 75%, 89% and 95% of trip duration). Results: In the majority of epochs (74%), child occupants' heads were observed to be 'optimally' positioned within the FFCRS or BS. For more than half of the epochs, child occupants were observed to be: correctly restrained (58%) and involved in an interaction with another vehicle occupant (59%). Bivariate analyses revealed that children travelling in a FFCRS were significantly more likely to be observed to have optimal head positions than those travelling in a BS (78% vs. 62%), χ2 (1) = 86.00, p < 0.001. Child occupants who were observed to be 'correctly' restrained were significantly more likely to be observed to have optimal head positions than those who were observed to be 'incorrectly' restrained (80% vs. 20%), χ2 (1) = 10.33, p < 0.01. Conclusions: This is the first naturalistic driving study (NDS) to specifically explore the factors associated with child occupants' head position when travelling in a CRS. Findings from the current study can be used to inform the positioning of anthropometric test dummies (ATD) in CRS testing, guide improvements to CRS/vehicle design, and develop targeted educational strategies to improve child occupant safety.

Standardized error severity score (ESS) ratings to quantify risk associated with child restraint system (CRS) and booster seat misuse.

OBJECTIVE: Although numerous research studies have reported high levels of error and misuse of child restraint systems (CRS) and booster seats in experimental and real-world scenarios, conclusions are limited because they provide little information regarding which installation issues pose the highest risk and thus should be targeted for change. Beneficial to legislating bodies and researchers alike would be a standardized, globally relevant assessment of the potential injury risk associated with more common forms of CRS and booster seat misuse, which could be applied with observed error frequency-for example, in car seat clinics or during prototype user testing-to better identify and characterize the installation issues of greatest risk to safety. METHODS: A group of 8 leading world experts in CRS and injury biomechanics, who were members of an international child safety project, estimated the potential injury severity associated with common forms of CRS and booster seat misuse. These injury risk error severity score (ESS) ratings were compiled and compared to scores from previous research that had used a similar procedure but with fewer respondents. To illustrate their application, and as part of a larger study examining CRS and booster seat labeling requirements, the new standardized ESS ratings were applied to objective installation performance data from 26 adult participants who installed a convertible (rear- vs. forward-facing) CRS and booster seat in a vehicle, and a child test dummy in the CRS and booster seat, using labels that only just met minimal regulatory requirements. The outcome measure, the risk priority number (RPN), represented the composite scores of injury risk and observed installation error frequency. RESULTS: Variability within the sample of ESS ratings in the present study was smaller than that generated in previous studies, indicating better agreement among experts on what constituted injury risk. Application of the new standardized ESS ratings to installation performance data revealed several areas of misuse of the CRS/booster seat associated with high potential injury risk. CONCLUSIONS: Collectively, findings indicate that standardized ESS ratings are useful for estimating injury risk potential associated with real-world CRS and booster seat installation errors.

The effect of driver eye height on speed choice, lane-keeping, and car-following behavior: results of two driving simulator studies.

OBJECTIVE: Two simulator studies were conducted that assessed the effect of driver eye height on speed choice, lane-keeping, and car-following behavior. The effect of eye height on the subjective variables of mental workload, frustration, and confidence was also investigated, as was the contribution of drivers' aggression. METHODS: A total of 43 participants drove a simulated route while seated at two different eye heights: one that represented the view of the road from a large SUV and one that represented the view of the road from a small sports car. Driving scenarios were comprised of both open road and car-following segments. Dependent variables included driver-selected speed, speed variability, lane position, following distance to a slower-moving lead vehicle, and the subjective variables of frustration, confidence, and mental workload. RESULTS: When viewing the road from a high eye height, drivers drove faster, with more variability, and were less able to maintain a consistent position within the lane than when viewing the road from a low eye height. Driver eye height did not influence following distance to a slower-moving lead vehicle. Driver aggression had no effect on any of the dependent variables except level of frustration. CONCLUSIONS: The two studies demonstrate that, when they are not able to reference a speedometer, drivers choose to drive faster when they view the road from an eye height that is representative of a large SUV compared to that of a small sports car. There is a need to educate drivers of SUVs and other tall vehicles of this perceptual phenomenon in order to prevent collisions that may occur in conditions where it is impossible for drivers to base their speed selection solely on posted speed limits, such as in inclement weather.

Evaluation of a video-based measure of driver heart rate.

INTRODUCTION: Internal driver events such as emotional arousal do not consistently elicit observable behaviors. However, heart rate (HR) offers promise as a surrogate measure for predicting these states in drivers. Imaging photoplethysmography (IPPG) can measure HR from face video recorded in static, indoor settings, but has yet to be examined in an in-vehicle driving environment. METHODS: Participants (N=10) completed an on-road driving task whilst wearing a commercial, chest-strap style heart rate monitor ("baseline"). IPPG was applied to driver face video to estimate HR and the two measures of HR were compared. RESULTS: For 4 of 10 participants, IPPG produced a valid HR signal (±5 BPM of baseline) between 48 and 75% of trip duration. For the remaining participants, IPPG accuracy was poor (<20%). CONCLUSIONS: In-vehicle IPPG is achievable, but significant challenges remain. PRACTICAL APPLICATIONS: The relationship between IPPG accuracy and various confounding factors was quantified for future refinement.

Usability issues concerning child restraint system harness design.

A study was conducted to assess usability issues relating to child restraint system (CRS) harness design. Four convertible child restraint systems representing a wide variety of design features were used. Forty-two participants installed two child test dummies in both forward- and rear-facing configurations either inside or outside a test vehicle. Observer-scored checklists determined the degree to which each harness was installed correctly. Participant-scored questionnaires evaluated the 'ease-of-use' of various design features. While the percentage of correct installations exceeded 83% for all designs when installed in the forward-facing configuration, in the rear-facing position (that intended for children under 9-10 kg), there was a significant (between 65 and 89%) percentage of incorrect installations for all models. This finding is of particular interest and may be indicative of a more generalized problem with 'convertible' CRS designs when they are used in the rear-facing configuration. Furthermore, while certain design features were perceived by users as providing significantly better protection in the event of a collision, these also tended to be the features that were misused most often. The benefits and costs of various design features are discussed, and a method to test harness design usability is presented.

The design of child restraint system (CRS) labels and warnings affects overall CRS usability.

A study was conducted that assessed the effectiveness of different child restraint system (CRS) label/warning designs on users' installation performance. Forty-eight paid participants installed a convertible CRS in a vehicle, and two child test dummies in a CRS, using one of four label conditions. The label conditions were: (1) no labels, (2) the manufacturer's labels that were already affixed to the CRS ("Current"), (3) labels that were designed according to a combination of the current U.S. regulations concerning CRS labels and recently proposed changes to these regulations ("Proposed"), and (4) labels that were designed according to human factors principles and guidelines, and that were based on a hierarchical behavioral task analysis ("Optimal"). Results demonstrated that, overall, the Optimal labels resulted in higher usability ratings and better task performance. This indicates that labels designed using human factors and task analyses that identify critical task information requirements for label features will result in increased user compliance with instructions, higher usability, and improved task performance. Surprisingly, having no labels on the CRS resulted in better installation performance than when either the Current or the Proposed label conditions were used. This indicates that label design can decrease task performance; the actual physical design of a CRS may be just as critical as label content in the installation choices provided to the user. Collectively, results suggest that implementation of the proposed changes to the U.S. regulations concerning CRS labeling would likely not result in increased performance or usability compared to existing manufacturer labels that follow the current guidelines. In order to achieve significantly better ease-of-use and task performance, it would be necessary to implement features of the Optimal label condition.

Computer vision and driver distraction: developing a behaviour-flagging protocol for naturalistic driving data.

Naturalistic driving studies (NDS) allow researchers to discreetly observe everyday, real-world driving to better understand the risk factors that contribute to hazardous situations. In particular, NDS designs provide high ecological validity in the study of driver distraction. With increasing dataset sizes, current best practice of manually reviewing videos to classify the occurrence of driving behaviours, including those that are indicative of distraction, is becoming increasingly impractical. Current statistical solutions underutilise available data and create further epistemic problems. Similarly, technical solutions such as eye-tracking often require dedicated hardware that is not readily accessible or feasible to use. A computer vision solution based on open-source software was developed and tested to improve the accuracy and speed of processing NDS video data for the purpose of quantifying the occurrence of driver distraction. Using classifier cascades, manually-reviewed video data from a previously published NDS was reanalysed and used as a benchmark of current best practice for performance comparison. Two software coding systems were developed - one based on hierarchical clustering (HC), and one based on gender differences (MF). Compared to manual video coding, HC achieved 86 percent concordance, 55 percent reduction in processing time, and classified an additional 69 percent of target behaviour not previously identified through manual review. MF achieved 67 percent concordance, a 75 percent reduction in processing time, and classified an additional 35 percent of target behaviour not identified through manual review. The findings highlight the improvements in processing speed and correctly classifying target behaviours achievable through the use of custom developed computer vision solutions. Suggestions for improved system performance and wider implementation are discussed.

A preliminary estimation of motorcyclist fatal injury risk by BAC level relative to car/van drivers.

OBJECTIVE: This study sought to quantify the fatal injury risks for motorcyclists associated with the riders' blood alcohol concentrations (BACs). METHOD: Using a case-control study design with New Zealand data, fatal injury risk curves for motorcyclists and car drivers were modeled. A total of 142 fatally injured drivers/riders (cases) and 58,000 control drivers/riders were studied. For motorcyclists, there were 13 cases and 194 controls. RESULTS: The rate of increase in fatal injury risk with increasing BAC was not found to be different for motorcyclists compared to car/van drivers. However, because the baseline risk for motorcyclists was already considerably higher than for car/van drivers, even modest amounts of alcohol were associated with very high risks for motorcyclists compared to sober car/van drivers. It was estimated that, relative to their sober risk, motorcycle riders at BAC = 0.03 percent have 3 times the fatality risk (95% confidence interval [CI] = 2.8-3.5) and, at BAC = 0.08, 20 times the fatality risk (95% CI = 15.0-27.3). CONCLUSIONS: Interventions focused on reducing the alcohol consumption of motorcycle riders are clearly required when the degree of risk even at low alcohol levels is as disturbingly high as estimated in the current study.

A simulator study of the effects of singing on driving performance.

This study aimed to investigate how singing while driving affects driver performance. Twenty-one participants completed three trials of a simulated drive concurrently while performing a peripheral detection task (PDT); each trial was conducted either without music, with participants listening to music, or with participants singing along to music. It was hypothesised that driving performance and PDT response times would be impaired, and that driver subjective workload ratings would be higher, when participants were singing to music compared to when there was no music or when participants were listening to music. As expected, singing while driving was rated as more mentally demanding, and resulted in slower and more variable speeds, than driving without music. Listening to music was associated with the slowest speeds overall, and fewer lane excursions than the no music condition. Interestingly, both music conditions were associated with slower speed-adjusted PDT response times and significantly less deviation within the lane than was driving without music. Collectively, results suggest that singing while driving alters driving performance and impairs hazard perception while at the same time increasing subjective mental workload. However, singing while driving does not appear to affect driving performance more than simply listening to music. Further, drivers' efforts to compensate for the increased mental workload associated with singing and listening to music by slowing down appear to be insufficient, as evidenced by relative increases in PDT response times in these two conditions compared to baseline.

Performance of a cognitive, but not visual, secondary task interacts with alcohol-induced balance impairment in novice and experienced motorcycle riders.

The appropriateness of applying drink driving legislation to motorcycle riding has been questioned as there may be fundamental differences in the effects of alcohol on driving and motorcycling. It has been suggested that alcohol may redirect riders' focus from higher-order cognitive skills such as cornering, judgement and hazard perception, to more physical skills such as maintaining balance. To test this hypothesis, the effects of low doses of alcohol on balance ability were investigated in a laboratory setting. The static balance of twenty experienced and twenty novice riders was measured while they performed either no secondary task, a visual (search) task, or a cognitive (arithmetic) task following the administration of alcohol (0%, 0.02%, and 0.05% BAC). Subjective ratings of intoxication and balance impairment increased in a dose-dependent manner in both novice and experienced motorcycle riders, while a BAC of 0.05%, but not 0.02%, was associated with impairments in static balance ability. This balance impairment was exacerbated when riders performed a cognitive, but not a visual, secondary task. Likewise, 0.05% BAC was associated with impairments in novice and experienced riders' performance of a cognitive, but not a visual, secondary task, suggesting that interactive processes underlie balance and cognitive task performance. There were no observed differences between novice vs. experienced riders on static balance and secondary task performance, either alone or in combination. Implications for road safety and future 'drink riding' policy considerations are discussed.

Driver distraction in an unusual environment: Effects of text-messaging in tunnels.

Text messaging while driving can be distracting and significantly increases the risk of being involved in a collision. Compared to freeway driving, driving in a tunnel environment introduces factors that may interact with driver attentional resources to exacerbate the effects of distraction on driving safety. With planning and design of the 18km Stockholm Bypass tunnel ongoing, and because of the potentially devastating consequences of crashes in long tunnels, it is critical to assess the effects of driver distraction in a tunnel environment. Twenty-four participants (25-50 years) drove in simulated highway and tunnel road environments while reading and writing text messages using their own mobile phones. As expected, compared to driving alone, text messaging was associated with decrements in driving performance and visual scanning behavior, and increases in subjective workload. Speeds were slower compared to baseline (no text-messaging) driving when participants performed the text-messaging tasks in the tunnel environment compared to the freeway, suggesting that drivers may have attempted to compensate more for the increased text-messaging-related workload when they were in the tunnel. On the other hand, increases in lane deviation associated with the most complex text-messaging task were more pronounced in the tunnel compared to on the freeway. Collectively, results imply that driver distraction in tunnels is associated with generally similar driving decrements as freeway driving; however, the potential consequences of these decrements in tunnels remain significantly more serious. Future research should attempt to elucidate the nature of any differential compensatory behavior in tunnel, compared to freeway, driving. In the meantime, drivers should be advised to refrain from text messaging, especially when driving in tunnels.

The effects of on-street parking and road environment visual complexity on travel speed and reaction time.

On-street parking is associated with elevated crash risk. It is not known how drivers' mental workload and behaviour in the presence of on-street parking contributes to, or fails to reduce, this increased crash risk. On-street parking tends to co-exist with visually complex streetscapes that may affect workload and crash risk in their own right. The present paper reports results from a driving simulator study examining the effects of on-street parking and road environment visual complexity on driver behaviour and surrogate measures of crash risk. Twenty-nine participants drove a simulated urban commercial and arterial route. Compared to sections with no parking bays or empty parking bays, in the presence of occupied parking bays drivers lowered their speed and shifted their lateral position towards roadway centre to compensate for the higher mental workload they reported experiencing. However, this compensation was not sufficient to reduce drivers' reaction time on a safety-relevant peripheral detection task or to an unexpected pedestrian hazard. Compared to the urban road environments, the less visually complex arterial road environment was associated with speeds that were closer to the posted limit, lower speed variability and lower workload ratings. These results support theoretical positions that proffer workload as a mediating variable of speed choice. However, drivers in this study did not modify their speed sufficiently to maintain safe hazard response times in complex environments with on-street parking. This inadequate speed compensation is likely to affect real world crash risk.

The implications of cross-regional differences for the design of In-vehicle Information Systems: a comparison of Australian and Chinese drivers.

The increasing global distribution of automobiles necessitates that the design of In-vehicle Information Systems (IVIS) is appropriate for the regions to which they are being exported. Differences between regions such as culture, environment and traffic context can influence the needs, usability and acceptance of IVIS. This paper describes two studies aimed at identifying regional differences in IVIS design needs and preferences across drivers from Australia and China to determine the impact of any differences on IVIS design. Using a questionnaire and interaction clinics, the influence of cultural values and driving patterns on drivers' preferences for, and comprehension of, surface- and interaction-level aspects of IVIS interfaces was explored. Similarities and differences were found between the two regional groups in terms of preferences for IVIS input control types and labels and in the comprehension of IVIS functions. Specifically, Chinese drivers preferred symbols and Chinese characters over English words and were less successful (compared to Australians) at comprehending English abbreviations, particularly for complex IVIS functions. Implications in terms of the current trend to introduce Western-styled interfaces into other regions with little or no adaptation are discussed.

The effects of using a portable music player on simulated driving performance and task-sharing strategies.

This study examined the effects of performing scrollable music selection tasks using a portable music player (iPod Touch™) on simulated driving performance and task-sharing strategies, as evidenced through eye glance behaviour and secondary task performance. A total of 37 drivers (18-48 yrs) completed the PC-based MUARC Driver Distraction Test (DDT) while performing music selection tasks on an iPod Touch. Drivers' eye glance behaviour was examined using faceLAB eye tracking equipment. Results revealed that performing music search tasks while driving increased the amount of time that drivers spent with their eyes off the roadway and decreased their ability to maintain a constant lane position and time headway from a lead vehicle. There was also evidence, however, that drivers attempted to regulate their behaviour when distracted by decreasing their speed and taking a large number of short glances towards the device. Overall, results suggest that performing music search tasks while driving is problematic and steps to prohibit this activity should be taken.

Effectiveness of traffic light vs. boom barrier controls at road-rail level crossings: a simulator study.

Although collisions at level crossings are relatively uncommon occurrences, the potential severity of their consequences make them a top priority among safety authorities. Twenty-five fully-licensed drivers aged between 20 and 50 years participated in a driving simulator study that compared the efficacy, and drivers' subjective perception, of two active level crossing traffic control devices: flashing lights with boom barriers and standard traffic lights. Because of its common usage in most states in Australia, a stop sign-controlled level crossing served as the passive referent. Although crossing violations were less likely at the level crossings controlled by active devices than at those controlled by stop signs, both kinds of active control were associated with a similar number of violations. Further, the majority (72%) of drivers reported preferring flashing lights to traffic lights. Collectively, results indicate that the installation of traffic lights at real-world level crossings would not be likely to offer safety benefits over and above those provided already by flashing lights with boom barriers. Furthermore, the high rate of violations at passively controlled crossings strongly supports the continued practice of upgrading level crossings with active traffic control devices.

Driver behaviour at rail level crossings: responses to flashing lights, traffic signals and stop signs in simulated rural driving.

Australian road and railway authorities have made a concerted effort to reduce the number of rail level crossings, particularly the higher risk passive crossings that are protected by devices such as 'give way' or 'stop' signs. To improve this situation, passive level crossings are often upgraded with active controls such as flashing red lights. Traffic signals may provide good safety outcomes at level crossings but remain untested. The primary purpose of this research was to compare driver behaviour at two railway level crossings with active controls, flashing red lights and traffic signals, to behaviour at the current standard passive level crossing control, a stop sign. Participants drove the MUARC advanced driving simulator for 30 min. During the simulated drive, participants were exposed to three level crossing scenarios. Each scenario consisted of one of three level crossing control types, and was associated with an oncoming train. Mean vehicle speed on approach to the level crossings decreased more rapidly in response to flashing lights than to traffic signals. While speed on approach was lowest for the stop-sign condition, the number of non-compliant drivers (i.e., those who did not stop) at the crossing was highest for this condition. While results indicate that traffic signals at rail level crossings do not appear to offer any safety benefits over and above flashing red lights, further avenues of research are proposed to reach more definitive conclusions. Compliance was lowest for the passive crossing control which provides further support for the ongoing passive crossing upgrades in Australia.