Enhanced fire escape training for mine workers using virtual reality simulation

Recent U.S. mine disasters identified a need for improved training of miners in escape from mine fires and explosions. Researchers at the National Institute for Occupational Safety and Health’s (NIOSH) Spokane Research Labora­ tory (SRL) have designed fire-escape training software for use in a mine-safety training course. The software allows four trainees to work together in a virtual world via a com­ puter network. Trainees responded favorably and showed a marked improvement in locating the proper evacuation routes on subsequent trials. In collaboration with the Pittsburgh Research Laboratory (PRL), a more elaborate fire evacuation scenario is being developed. Further study is needed to confirm that virtual training will translate to improved safety outcomes in the real world. Introduction U.S. Mine Safety and Health Administration (MSHA) accident reports often cite poor training as a root cause for many fatalities and serious injuries, but traditional classroom safety training provides little practical experience for mine workers. Virtual reality (VR) training systems have been used successfully in many dif­ ferent settings including pilot training, military simulators and mining equip­ ment operation. This type of training has the added benefit of providing simulated experience in dealing with a hazardous circumstance on the job without exposing the trainee to the hazard. Although this type of training has been successfully used in other fields, it has not been used much or tested for effectiveness in mining. Therefore, this study answers the following research question — can a VR system be used in a mine safety train­ ing environment to teach evacuation procedures, provide simulated experience, and measure performance? The study was an informal, preliminary test of the Vir­ tual Reality Miner Safety Training (VRMST) software, developed at the National Institute for Occupational Safety and Health’s (NIOSH) Spokane Research Labora­ tory (SRL). The VRMST software was designed for mine workers to practice evacuation routes and procedures. A preliminary test of this software was conducted during an eight-hour, MSHA-approved refresher training course. This training was led by an approved mine safety trainer and SRL research staff were the subjects of the prelimi­ nary testing of the VR training. Based on these initial results, a new project was funded to further develop and enhance the VRMST software into a publicly available product and determine how the soft­ ware could be distributed to safety trainers. The new simulation software uses the Unreal game engine from Epic Games (www.unrealtechnology.com). SRL researchers modified some of the high-level script code and included custom maps, models and animations. By repurposing the computer game in this fashion, the end user does not see any elements of the original game on the screen. The software provides a first-person perspective on the virtual environ­ ment — meaning the computer dis­ play renders the virtual environment from the view of the person sitting at the keyboard as opposed to a top-down, or third-person perspective. Numerous researchers have used the power of this game engine for nongaming pur­ poses such as teaching pedestrian safety (McComas, et al, 2002), creating architectural walkthroughs (Johns and Lowe, 2006) and military simulations (Ryan, et al, 2005; Wray, et al, 2004). In response to this, Epic Games decided to release a version of its game engine for nongaming, noncommercial purposes. The new NIOSH virtual reality training software is built on this game engine, which will allow the Institute to publish and distribute the resulting training products to the public. The first part of this paper describes an earlier software version that was never pub­ licly released but led to the development of the current software package scheduled for release in late 2008. The simulation software currently under development is de­ tailed at the end of this paper. Design of VR software The simulation software uses a computer network to place four trainees within the same simulated environment (Fig. 1). Each trainee is represented in the virtual mine by a computer-generated character called an avatar. Trainees have independent control over their avatar in the virtual mine using simple keyboard and mouse controls. The simu­ lation style is first-person, meaning that each trainee can see the other trainees’ avatars, but they do not see their own avatar on their computer screen (Fig. 2). First-person simulation attempts to provide a virtual “first-hand” per­ spective of the setting. The virtual mine for this training simulated a com­ mon setup in underground longwall coal mining with a three-entry gate road development heading. A three-entry development consists of three parallel entries connected by crosscuts. Mine workers construct stoppings in these crosscuts as mining progresses to control ventilation air flow routed through the entries. Mandoors may be placed in some stoppings to allow travel from one entry to the next. In actual work settings, there are limited options for escape and miners, particularly new employees, can become easily confused about which escape route to take (Murray, et al, 2007). The simulated evacuation scenario used in this train­ ing exercise was based on an actual incident involving a fire at a cooperating mine. The simulation included smoke that significantly obscures the trainees’ VR vision in some areas. In addition, the virtual environment included falls of ground (as could be caused by the fire) that blocked the trainees’ primary escape route and forced them to find an alternate path, just as they might experience in an actual evacuation. Trainees were also presented with VR obstacles that required them to consider various escape routes. This type of critical decision-making takes time and requires the trainees to analyze their virtual experience and talk with each other, via telephone if the computers are connected from remote locations. Figure 1 The VR training computer network used for the test.