Communication and mobility are a person's primary needs. Some people with Traumatic Brain Injury (TBI) face difficulties in performing complex visio spatial tasks such as driving a power wheelchair. Computer based virtual environments are inherently safe and can be customized for the user. Hence they are a preferred option for initial driving training. This paper discusses the development of such a 2D computer based simulation for training and evaluating the wheelchair driving performance using a prototype isometric joystick.
2D Wheelchair driving simulator, virtual reality, head position monitor, isometric joystick
Some people with traumatic brain injury (TBI) experience visual field and compromised attention spans . This makes the task of electric powered wheelchair (EPW) driving difficult and sometimes dangerous for them, especially if they don't have prior experience in driving. Independence in mobility has positive reinforcements towards improving the quality of life of people with disabilities  and hence steps should be taken to teach them to drive an EPW. Such training in real world would require supervision by the therapist to ensure safety of the user. A virtual environment is inherently safe for training purposes and virtual wheelchair driving performance has been shown to correlate well with the performance real world .
Highly immersive Virtual Reality (VR) environments have been developed  for this purpose but they are expensive to implement in a small clinical setting. Further, cyber sickness experienced due to such environments  and their operational complexity place limitations to their clinical use. These are some reasons computer based wheelchair driving simulators are developed in recent years.
Current research at Human Engineering Research Laboratories (HERL) is aimed to develop a clinically relevant computer based training environment, especially for people with limited cognitive abilities for training EPW driving using an isometric joystick.
The virtual wheelchair driving software (VWDS) runs on the operator's computer with a large television screen (27") display. The user sits in a stationary wheelchair and operates the simulation with the isometric joystick. He/she wears a hat with magnet attached to it posteriorly and rests his/her head on the headrest shaped head position monitor (HPM). The HPM has strategically mounted Hall Effect transistors that aid in detecting orientation of user's head by magnetic triangulation .
Presently the VWDS developed at HERL is a non immersive 2D simulation. The user operates the joystick to move a wheelchair sprite on a stationary track seen from a bird's eye viewpoint (Figure 1). Apart from the rectangular track 4 other tracks may be selected for evaluating turns and straight line motion. The backend software handles joystick input processing and mainly consists of following four modules. The flow of control between them is shown in figure 2.
FDA guidelines  are followed during the ongoing development and validation of the simulation software. Structural testing was performed to detect "dead code" that is never executed and that code was replaced. Software specifications for the user interface and implementation of specialized algorithms have been implemented and are currently in the process of validation.
The physical construction of isometric joystick does not allow for features like template (constraints to joystick motion during application of maximum force) and dead zones. These features are implemented through software as they play an important role in tremor suppression and make the wheelchair sprite appear to be stable (simulating inertia of actual wheelchair). The bias axis and gain of joystick may be adjusted in software for better tuning the joystick according to the user's needs.
The present simulation faces some limitations in portability across systems and in implementation of contact detection algorithm. Other than halting wheelchair if the direction of visual focus and head orientation are different currently there is no other feedback provided to the user.
Stringent verification and validation techniques will be implemented to make wheelchair motion in simulation closer to real world driving. The presently developed 2D interface has advantages due to its simplicity in implementing and validating real world wheelchair behavior. A slightly more immersive computer based 3D environment may be a next step in the development process after adequate confidence from the 2D interface.
This research is supported by the University of Pittsburgh Model Center for Traumatic Brain Injury (TBI) Model Systems Grant H133A020502.
Human Engineering Research Laboratories
VA Pittsburgh Healthcare System
151R1, 7180 Highland Drive
Pittsburgh, PA 15206.
This should be in the right column.