Efficacy of Enhancing Low Vision Mobility Thru Visual Training in Virtual World

NEI's National Plan for Eye and Vision Research has identified Low Vision and Blindness Rehabilitation as one of the most pressing needs and opportunities in vision research. Since most low vision (LV) patients nowadays have useable vision, an important aspect of LV rehabilitation is to maximize the use of the patient's remaining vision. Visual skills are especially important in tasks such as traveling independently through busy streets, in which distant, rapidly changing events have direct impact on the patient's task efficiency and safety. The patient has to learn new information acquisition skills to compensate for their impaired vision. Current LV training for such skills is done on real streets in a one-on-one, trainer/trainee sessions. This way of training, however, usually does not expose clients to all the circumstances that they need to deal with, usually does not provide graded levels of task difficulty and repeatable conditions that facilitate learning, and cannot meet the needs of the majority of LV patients because of the shortage and the high cost of LV trainers. The advances of virtual reality (VR) technology have made it possible and affordable to realistically simulate complex and dynamic visual and auditory environments to facilitate LV rehabilitation. However, the most important first step in applying VR technologies to LV rehabilitation is to determine whether the visual experiences of a LV person can be transferred to the real world. Only when a positive transfer of visual skills is confirmed, can subsequent development of comprehensive VR training systems for all aspects of LV rehabilitation become worthwhile. The purpose of this study is to accomplish this first step. A set of visual skills for a most visually challenging task, crossing a signal-controlled street, is selected from standard orientation and mobility curriculum to study virtual experience transfer. Existing technology are used to build a desktop computer-based VR simulator that can generate virtual street intersection scenarios that contain visual and auditory information required for practicing the skills to be studied. Computer generated street intersections with pedestrians, traffic and control signals are projected to a large screen for the participant to view, like a movie. Twenty LV persons are recruited from a university LV rehabilitation facility. They are randomly assigned to practicing the selected visual skills at real or virtual street intersections. Their skill levels are evaluated at real intersections before and immediately after their scheduled training. The primary outcome of interest is the mean change of visual skills between pre- and post-training evaluations in the two groups. The training is not a rehabilitation intervention because only the changes in visual/audio information gathering skills are trained and observed. The participants are trained and evaluated on the sidewalks of the real and virtual street. They are never allowed to step down the curb to avoid exposure to any danger. The long-term goal is to develop interactive VR training systems that have multiple sensory and motor input and output modules and that incorporate the collective experiences of the LV rehabilitation profession. Adding such systems to LV curriculum can greatly improve the affordability, accessibility, efficiency and the scope of LV rehabilitation.