Train-vehicle/train collisions at highway-rail grade crossings (crossings) continue to be a major issue in the US and around the world. Although the United States has made great strides in improving safety at crossings since the 1970s, vehicle-train accidents remain a major concern for the rail industry, costing millions of dollars every year and taking or destroying countless lives.

Noncompliant behavior such as gate-running is the biggest cause of accidents at crossings protected by active warning devices, while failure to detect the crossing or train, or at least, a poor time-to-arrival judgment, is the largest cause of accidents at crossings protected by passive warnings. Crossings with active warnings have lower crash risks per volume of vehicle traffic than those with passive warning devices. However, the cost to install and maintain active warnings is considerable and likely a barrier in many areas. Therefore, it is not only important to better understand the reasons behind driver’s non-compliant behavior, but also to assess the effectiveness and potential safety benefits of alternative warning approaches.

Because the nature of the breakdown in safety is due to different issues depending on the type of crossing, multiple countermeasures to increase safety must be developed to specifically target each issue independently and to promote respect for crossing warnings in general. In order to better understand how and why these collisions occur, we conducted a number of studies that investigated the psychology behind driver decision making on approach to crossings in a medium fidelity driving simulator.

The empirical investigation was split into two separate phases. The first phase of studies focused on the difference in driver response between different warning types. In Phase two of the driver behavior studies, In-Vehicle Auditory Alerts (IVAA’s) were developed and tested. Now that GPS, smartphone, and in-vehicle display technology have become less expensive and more popular, In-vehicle Auditory Alerts (IVAAs) for crossings are a potential low cost intervention that could upgrade all crossings nationwide at once. As of 2015, the FRA and Google have entered into a partnership to include the location of all RR crossings in Google Maps services, although no formal plans have been made public to introduce in-vehicle warnings for crossings. A type of warning system that could be easily implemented with the current technology is to provide in-vehicle notifications to alert the driver to the presence of crossings. This was selected as the objective of phase two. In this

study a pool of auditory cues was created and evaluated. Based on the results we designed and tested a prototype IVAA system.

Approach and Methodology

Phase 1

This was our initial foray into using the simulator to study crossing behavior. Working with the National Advanced Simulator (NADS) scenario development team from Iowa State University we developed a set of prototype crossing scenarios. As we began testing with these scenarios we quickly identified some problems, including an active crossing system with a gate but no lights, and missing advanced warning signs. As building new scenarios takes time, we elected to work with the existing scenario package through our first study to gain an understanding of the system and its strengths and weaknesses. We also began updating scenarios for use in later research efforts.

Compliance was measured via eye trackers (visual scanning behavior), and vehicle speed collected from the driving simulator software. The virtual environment featured an open rural road with high visibility. Participants drove for twenty minutes and experienced one of three warning types (a full gate with lights, a gate without lights1, and crossbuck only) approximately once every 5 minutes. A critical zone just before passing the crossing was defined for later comparison to non-critical zones before and after the crossing. These zones were analyzed as 8, 10, and 20 second intervals. Variables such as vehicle speed and eye tracking data were compared before during and after the critical zones. Each participant were exposed to all three of the warning device types. The order of the warning device type was counterbalanced across all participants. Before the experiment, a simulation sickness test was run during a practice driving session to identify and remove participants that exhibited any type of simulator sickness. The eye tracker was configured based on each individual participant’s height, distance, and angle from the cameras. Half of the participants were assigned to a condition where a train was present in the first crossing. The other half of the participants were assigned to a control condition with no train presence. Participants were recruited from undergraduate psychology classes from the Michigan Technological University SONA system. Participants were rewarded with class credit in exchange for 1 hour of their time. Participants were only informed that the experiment was observing driving behavior in a driving simulator, and were unaware about the possibility of simulated trains or rail road crossings.