Ⅰ. Introduction

This study aims to evaluate driver reactions to the introduction of new ITS interventions at railway crossings (RLX) in driving simulator. Systems which complement standard controls at RLXs are not expected to have the same level of integrity (SLI), but could be used by drivers as a primary control; however, these would have negative outcomes in terms of safety at the crossing. Therefore, studying the effects of such devices must first be performed in a controlled environment, which can be provided by a driving simulator. A little research has been conducted on safety at RLXs [1,2] compared to road intersections since the number of collisions at RLXs is significantly less than road intersections. Even though a few research considered on general safety at RLXs, taking ITS devices into account safety is hardly seen.

It implements the interfaces with the highest potential to increase safety at RLXs. A mock-up of the most suitable technology is simulated and tested in an advanced driving simulator. Using an advanced driving simulator has the following advantages:

Scenarios can be analysed with experimental repeatability, easy configuration ability and excellent data collection capability. This facilitates case control, before and after studies. Finally, it allows quick and inexpensive ways to test many non-existent road safety interventions for railway crossings before trialling the most promising at a real RLX.

However, some of the drawbacks are simulator [3,4], whereby a participant experiences symptoms of motion sickness even after only brief exposure to the driving simulator, due to the lack, or incomplete replication of physical sensations [3]; no possibility to use multiple drivers interacting while driving; and validity of the measures provided by the simulator.

This study is a simulated driving task experiment. It aims to assess the effects of various in-vehicle and on-road ITS interventions on driver behaviour at both active and passive railway level crossings. Three different ITS interventions are trialled, as well as a baseline of standard active and passive crossings.

This study investigates the following three ITS interventions:

The objective of this study is to evaluate the safety impacts of these different assistive systems on human errors, intentional actions, and objective and subjective risk assessments. This study evaluates the resulting objective and subjective behavioural changes in different traffic situations which are modelled to be representative of the different types of crossings (active and passive).

This study used questionnaires for assessing subjective feedback from participants primarily assesses the effects of such interventions on driver decisions and behaviour at railway crossings through:

Ⅱ. Experiments

1. Participants’ demographics

80 participants were recruited to participate in this study. 23 participants were not able to complete the three drives (runs) and were not considered during the analysis. This study has therefore a sample size N=57, composed of 39 males and 18 females. The average age is 28.2 years, with a standard deviation of 7.63. Ages ranged from 19 to 59. Participants were divided into three groups, each group trialling one particular ITS intervention. The first group trialled the visual in vehicle ITS and was composed of 18 participants. The second group trialled the audio in vehicle ITS and was composed of 22 participants. The last group trialled the on-road valet system and was composed of 17 participants.

2. Design scenarios

Participants drive in the simulator on two types of crossings, namely passive and active with flashing lights only. Participants are randomly divided into three groups (each group is characterised by an ITS intervention) and drive without any ITS device (baseline) and with one type of ITS device per group. This means a within-subject design is selected in order to increase the power to detect differences between the baseline and ITS interventions. There can be carry- over (or contamination) effects from one experimental condition to the next when a within-subject design is used. This would dilute the between-condition differences of the outcomes, possibly offsetting the benefits gained from using the within-subject design. In this setting, using different ITS after each other could create such contamination [5]. It has therefore been decided that each participant uses only one type of ITS in order to remove any potential carry- over effects which could result from driving with two different interventions one after the other.

3. Driving simulator

Experimentation is conducted on an advanced driving simulator. This simulator is composed of a complete automatic Holden Commodore vehicle with working controls and instruments. The advanced driving simulator housed in Centre for Accident Research and Road Safety, Australia uses SCANeR studio software with eight computers, projectors and a six degree of freedom (6DOF) motion platform that can move and twist in three dimensions. When seated in the simulator vehicle, the driver is immersed in a virtual environment which includes a 180 degree front field of view composed of three screens, simulated rear view mirror images on LCD screens, surround sound for engine and environment noise, real car cabin and simulated vehicle motion (see Fig. 1). The road and environment are developed to respect Standards at RLXs, as well as create realistic traffic around the driven car. The rendering capabilities of this simulator enable it to display a realistic driving environment (highway, country road, city). In order to produce a realistic scenario, SCANeR can be programmed to control all the elements of the environment (point of view, vehicles, trains, pedestrians, trees, buildings, road signs, lights, etc.) in advance or after specific events.

SCANeR also enables us to create realistic road networks with a combination of straight and curvy sections with their intersections. It is possible to use satellite pictures or road maps to design the network and altitude can be modelled in the second step. SCANeR also contains a database of static environments such as buildings, signages, cars, pedestrians, trees and ground textures.

The participant sits in the driver’s seat of the car. They can see three screens where the SCANeR simulation is played by three RGB video projectors. The participant drives the simulator with two pedals (brake and accelerator only) and a steering wheel which provides force feedback.

4. Signage Standards

Signage at crossing will be designed to follow standards. In this experiment, passive crossings and active crossings without barriers will be implemented. The first three types of crossings follow standards and details of their geometry and required signage are detailed in Fig. 2.

5. ITS safety devices

Three different ITS will be implemented in this experiment and their mock-up will be presented in the following subsections. They will provide similar information through different HMIs.

1) Visual ITS (ITS1)

The visual in-vehicle ITS is implemented with a smart phone. This smart phone is positioned within the driving cabin at the usual location of a GPS. Fig. 3 shows how this device is integrated in simulator, in the middle of the dashboard next to the two cameras of the eye tracker device.

As a train is approaching the crossing, the smart phone will display a warning flashing picture synchronised with the flashing lights of active crossings. For passive crossings the warning will be displayed at an equivalent time. In this situation (train approaching), the warning will provide two messages at the same time in one symbolic representation: the fact that a train is approaching the crossing and that the driver is expected to stop.

The rendering of the implementation of the visual in-road ITS is presented in Fig. 4, which provide pictures of the smart phone as it displays a warning message.

2) Audio ITS (ITS2)

The audio in-vehicle ITS uses the speakers of the simulator positioned inside the car (under the seat) to provide warning messages to the driver.

As a train is approaching the crossing, the speakers will provide a verbal warning as the flashing lights of active crossings are activated. For passive crossings the warning will be provided at an equivalent time. In this situation (train approaching), two messages will be given to the driver as in the visual ITS presented before:

Such system should improve driver awareness as the two previous ITS implementations presented earlier. This ITS is fully implemented with the simulator, as the speakers of the simulator will be used. Through scripting in the simulator, the messages can be played as the status of the crossing changes and requires a particular warning.

3) On-road flashing markers (ITS3)

This road-based ITS uses flashing warning beacons on the road which are activated when a train is approaching the crossing. These beacons highlight, in a similar way as illuminated airplane runways, the location where the driver is expected to stop their vehicle. This system improves driver’s awareness of the crossing status earlier and more conspicuously, even in the case of reduced visibility independent of the time of the day. This system should primarily provide benefits when the approach to the crossing is curved or inclines, and in foggy or sun glare conditions.

Flashing markers on the road are activated at the same time as the flashing lights of an active crossing and are positioned up to 150 metres from the crossing. In the case of passive crossings, the lights will be activated 20 seconds prior the arrival of the train, which provides a similar time to the driver to react to the warning. Three in-road red lights will be used to emphasise the stop line at the crossing. Five in-road yellow lights will be positioned in the middle of the road every 6 metres, and a further ten in-road yellow lights will be positioned every 12 metres.

Each individual flashing beacon is designed following Standards reflective road markers. A light is then added to the 3D object in order to make it flash as it is activated. The size of such reflective road markers is 10x10x2 cm. A halo is also added around the marker when it is activated in order to reinforce the visibility of such intervention in the simulator. The rendering of the implementation of the on-road ITS is presented in Fig. 5, which provides screen captures of the simulator from the driver’s view.

Ⅲ. Results - Questionnaires

1. General perception to crossings

Two types of crossings were tested; a passive crossing and an active crossing without boom barriers. Passive crossings are crossings which do not give information regarding the arrival of a train (static signs only: presence of railway crossings and stop sign/give way sign). Active crossings are crossings with lights that warn of the arrival of a train.

Participants were asked questions in relation to their perceptions to current railway crossings before they started doing experiment in driving simulator. 7 point scale was adopted to measure the perception to crossings. It represents extremely negative, quite negative, slightly negative, undecided, slightly positive, quite positive, and extremely positive for -3, -2, -1, 0, 1, 2, and 3 respectively).

Four questions were asked for participants as following;

If you had to drive though a passive/active crossing like in Fig. 6

As shown in Fig. 7 drivers feel ‘Quite unsafe’ (M: -1.9, SD: 0.27) and ‘Quite hard’ (M:-1.9, SD: 0.23) in Q1 and Q2 respectively to encounter passive crossings. They think that active crossings are not as much dangerous as passive crossings. However, they still reckon passing active crossings is ‘Slightly hard’ (M:-0.3, SD: 0.11). Accordingly, participants have a general idea that these two crossings are ‘Slightly prone to accident’ for themselves as well as other drivers. Basically, drivers do not feel safe or easy when they approach railway crossings without boom barriers regardless of a type of the crossing. Also, they have a thought that more collisions might occur at passive crossings. Some drivers mentioned that although an active crossing is equipped with flashing lights, they had lack of confidence due to malfunctioning safety devices. Consequently, the easiness to cross the active crossing is still quite negative without a physical obstacle like boom barriers.

2. Driver workload (NASA-TLX)

The NASA-TLX is administered to participants at the end of each drive, in order to obtain their subjective assessment of the mental and physical workload required for the drive (with or without the ITS).

A common measure of interest in driving simulation is the amount of workload Driver workload refers to the amount of effort a driver devotes to the driving task. It has also been defined in general as a set of task demands, as effort, and as activity or accomplishment [6] where the task demands are the goal to be achieved, including the time allowed to perform the task, and the performance level to which the task is to be completed [7]. Workload is a multidimensional construct involving interactions between the task and system demands, the operator (including mental and emotional capabilities) and the environment ([8]. In the driving context, workload is commonly defined as the effort required to maintain the driving state within a subjective safety zone [9]. In transportation research, driver workload can also refer to the amount of effort a driver devotes to the driving task.

One of the most commonly used subjective workload questionnaires used in driving research is the NASA Task Load Index, or NASA-TLX [10]. The NASA TLX is a multidimensional rating instrument that assesses six dimensions of subjective workload: mental demand, physical demand, temporal demand, performance, effort, and frustration level. Participants are required to indicate their subjective experience of workload in each of these six categories by indicating a point on a graded scale. There are also subjective workload measurement scales which have been specifically designed to assess driver workload [11].

Hart and Staveland’s NASA Task Load Index (TLX) method assesses work load on five 7-point scales. Increments of high, medium and low estimates for each point result in 21 gradations on the scales.

Table 1 presents the questions and their descriptions by adopting NASA-TLX.

<Fig. 8> shows results of NASA-TLX questionnaire. Interpretation of each question is as follows;

∙Mental demand (Q1)

Bar graphs of the NASA-TLX results for mental demand are presented in Fig. 8. It shows the level of mental demand between the different ITS technologies, as subjectively assessed by participants between 0 and 20. The reported value of mental demand by participants is 7.9 (SD: 6.0), 9.9 (SD: 5.9), and 7.3 (SD: 5.0) for ITS1, ITS2 and ITS3 respectively. ITS2 requires more mental demand for drivers than others.

∙Physical demand (Q2)

Drivers felt less physical demand than mental demand for all ITS interventions. Among ITS interventions, ITS1 requires less physical demand (M: 4.9, SD: 5.0) than others (M: 6.4, SD: 4.5) for ITS2 and M: 6.0, SD: 3.8 for ITS3).

∙Temporal demand (Q3)

Participant shows temporal demand corresponds to physical demand as Fig. 8 shows a similar pattern each other. The less temporal demand the less physical demand. Drivers did not hurry when their physical demand was not significant.

∙Effort (Q4)

Drivers made the biggest effort when they encountered crossings with ITS 2 whereas they completed relatively easier with ITS 1. Results of this index derive from a combination of mental, physical demand, temporal demand.

3. Measurement of driver acceptance of the technology

At the end of the last drive, a questionnaire about the acceptance of the particular ITS technology they were using in the simulator is administered to the participant. The knowledge of road rules at railway crossing of the participants is also assessed at the end of the drive, since many studies have highlighted the poor understanding of these rules [12,13].

It is important to highlight that approaches utilising emerging technologies are not to be viewed as a substitute for more traditional approaches to railway level crossing safety. To date, emerging technologies have typically been developed to target only one of the two goals of railway level crossing safety systems: that being improved detection of crossings and trains. That is, they largely fail to address the need to eliminate the ability of the driver to circumvent the technology. Indeed, short of immobilising the vehicle prior to a level crossing in anticipation of an impending collision (which involves its own safety risks), most emerging technology approaches can easily be bypassed or ignored by the driver. Thus, such approaches must be seen as complementary to existing traditional approaches.

To ensure compliance, it is necessary that there are high levels of driver acceptability of railway level crossing warning devices, such that motorists perceive the system to be reliable and understand the need for the device [14]. Driver acceptability is largely contingent on the perceived credibility of the warning device.

ITS interventions are efficient in reducing crashes at RLXs only if such technologies are deemed to be acceptable by drivers. Indeed, if such technologies do not respond to drivers’ needs and expectations, they are unlikely to be purchased or switched on, change driver behaviour and hence have positive effects on safety [15]. The design of such technologies must be centred on the user rather than technology driven, in order to ensure acceptability. That is why it is important to assess whether the different ITS devices tested in this project are likely to be accepted by drivers.

User acceptance – as defined by Swanson [16] – is a ‘potential user’s predisposition toward personally using a specific system’. Research supports a strong causal relationship between the behavioural intention to use a new technology and the actual behaviour [17]. The technology acceptance model (TAM) and the theory of planned behaviour (TPB) provide important theoretical bases in addressing the issue of users' adoption of new technologies [18].

Therefore, when using a pre-implementation perspective (the case of this study), it is necessary to have an experimental design in which the subjects were introduced and trained to use a new technology before they provide responses regarding the perceived usefulness, perceived ease of use and behavioural intentions toward that particular technology [19].

As a subjective prospective, it is of importance to Fig. out how drivers think of this technology and if they are willing to use this technology. For this information, following questions were asked to participants.

For the behavioural intention construct, as for the other constructs, the participants provided a level of intention to use the technology on a 7 scales. It represents strongly disagree, disagree, disagree somewhat undecided, agree somewhat, agree, strongly agree for -3, -2, -1, 0, 1, 2, and 3 respectively). Mean values of awareness, usefulness, easiness, and perceived usefulness are shown in Fig. 9 for the different type of ITS interventions and RLX protections.

Generally, ITS interventions are more attractive to drivers at passive crossings than active crossings.

∙Awareness (Q1)

Participants agreed that technology improved their awareness at the crossings as all cases for both passive and active crossings show about ‘Agree somewhat’ except for ITS1 at passive crossings indicating close to ‘Agree’ (M:1.06, SD:0.72). The technology helps driver’s awareness more at passive crossing than active crossings.

∙Perceived usefulness (Q2)

The Fig. 9 above shows that the perceived usefulness for ITS1 is nearly in ‘Agree’ category (M:1.56, SD: 0.46) while other ITSs are in almost ‘Agree somewhat’ category at the passive crossings. On the other hand, the Fig. 9 below shows that the perceived usefulness for ITS1 is in halfway (M: 1.89, SD: 0.63) between the ‘Agree somewhat’ and ‘Agree’ category. It is higher than other ITS interventions (ITS2: ‘Agree somewhat’ (M: 1.23, SD: 0.13), ITS3:’ between ‘Undecided’ and ‘Agree somewhat’ category (M: 1.18, SD: 0.20)) at the active crossings.

∙Perceived ease of use (Q3)

Perceived ease of use is a little above the ‘Agree’ level(M: 1.78, SD: 0.75) for ITS1 while it is slightly above ‘Agree somewhat’ level (M: 1.05, SD: 0.15) for ITS 2. This means that participants feel that ITS1 is much easier to use than any other ITS. This tendency follows at active crossings.

∙Behavioural intention (Q4)

This corresponds to the above ‘Agree’ category for ITS1 (M: 1.83, SD: 0.64) at passive crossing. The intention is followed by ITS 2 (between ‘Agree somewhat’ and ‘Agree’; M: 1.41, SD: 0.3) and ITS3( a little below ‘Agree somewhat’; M: 0.65, SD: 0.24). The intention to use these technologies at active crossings also shows similar level although the values are a little lower than those at passive crossings.

Ⅳ. Discussion and Future study

New Intelligent Transport Systems (ITS) based RLX interventions are now emerging due to the availability and the affordability of technology.

The behavioural intention to use the technology is quite positive, with a level slightly higher than the ‘Agree somewhat’ category. This value is a quarter of a level higher for passive crossings. For passive crossings, participants receive new information about the presence of trains and traffic. Participants tend to prefer in-vehicle interventions compared to the on road valet system. This suggests that participants might prefer an intervention at an in-vehicle intervention, rather than at the crossing itself.

The effect of type of protection at the crossing is not very large. This suggests that participants gave positive feedback for both active and passive crossings, with slightly higher values for passive crossings, where they gain more information. Audio and visual interventions have similar assessments.

Overall the three different ITS interventions do not create high workload to the driver. Mental, physical, temporal demands are slightly higher for the audio in-vehicle intervention, with 9.9, 6.4, and 7.5 respectively on a 21 point scale. Such a value could be due to the fact that the drivers do not visually see any information regarding train coming. While the in-vehicle video intervention shows the least demand to be used. Such a difference should not be an issue for the ITS intervention, but this result suggests that participants feel more comfortable with an intervention directly on the road or a visual intervention rather than an intervention that does not provide them with visual activity. As for the workload of participants, which was largely not changed, it appears that participants did not feel that they were performing either better or worse with the technology implemented at the RLXs.

Participants reported high intention to use the technology. Intention is a quarter of a level higher for passive crossings (on a 7 point scale). This could be due to the fact that at passive crossings, participants get new information about the presence of trains and traffic. Only the information about congestion at the crossing is new for active crossings, as the information about the train presence is only reinforcement.

This suggests that participants gave positive feedback for both active and passive crossings, with slightly higher values for passive crossings, where they gain more information. Overall, participants tend to favour the visual system compared to audio intervention.

This study should be followed by on-going research that embraces high-end technologies to help improve drivers’ awareness to warning signs. In addition to this, it is certainly important to develop an enhanced user interface that leads drivers to understand what new signs mean and what actions they need to do.