Virtual Reality Based Testing of Power Wheelchair Driving Skills

This study has an experimental, repeated measures design with validity and reliability testing. In Phase 1 of the study, we designed and developed the software needed to deliver the computer-based and VR-based testing environments. In Phase 2 of the study, we improved on the design iterations of Phase I, recruited a sample of 31 Electric Power wheelchair (EPW) users and delivered the simulator based driving assessment using a pool of clinicians to assess the reliability of system in real world and in the virtual environments.

Phase II: Research Protocol:

Experimental Set-up:

VRSIM consisted of two display options, three 6' X 8' back projected screens (field of view 160°) (VR screens) and a single 22" desktop monitor (field of view 90°) (PC screens), both with the first-person perspective viewpoint (Fig.1). The virtual environment consisted of a simulation of an indoor office space with a kitchen; a lounge area and set of hallways lined by offices, and incorporated the tasks of the PMRT (Fig.2). Participants interacted with VRSIM either using dual rollers that interfaced with drive wheels of the wheelchair ('Rollers ON' driving mode) or using an instrumented wheelchair joystick through a custom software ('Rollers OFF' driving mode). The custom software uses a proportional derivative mathematical model to simulate the real-world motion of the EPW within the virtual environment. VRSIM was designed with two display options and two user input modalities to assess the feasibility of using such a system with different interfaces in different settings, such as in a busy wheelchair clinic (using the much immersive VR screens with rollers) or in a user's home (with the user's personal computer and the customized joystick).

The "actor/driver" in the VRSIM was a virtual model of a person sitting in a standard commercially available EPW power wheelchair (width 0.671 m, length 0.701 m). A horizontal slider bar that indicated the real time location of the virtual wheelchair with respect to virtual obstacles was added to this version (Fig.1, 2). This helped display warnings when the chair was too close to obstacles, especially when the obstacles were behind the driver. Participants were expected to drive along the course indicated by arrows touching or passing through preset milestone markers signified by semi-transparent balloons. These sequentially displayed milestones defined the tasks of the virtual PMRT. Participants were instructed to complete every task as quickly and accurately as possible. An equivalent PMRT driving course was charted out in an office space for the real world driving evaluation.

Data collection:

After informed consent, participants performed up to 2 practice sessions within the virtual driving course. Participants reported their level of comfort and sense of being in control in the VRSIM during practice based on which a value between 1.0 and 4.0 was selected for the linear and angular speed gains in VRSIM. In addition, participants were asked to select a different pre-programmed driving profile on their wheelchairs (e.g. "indoor" profile), which was similar to their everyday driving profile to obtain optimum driving speed. Optional breaks for 5-10 minutes were provided between driving sessions. For both the PC and VR screens, participants drove through the complete driving course: two trials with the Rollers ON driving mode, one trial with the Rollers OFF driving mode, and one trial in the real-world PMRT driving course constituted to a total of 5 driving conditions (Table 1). A balanced randomization scheme was used to set the order of the five driving conditions.

From a group of 6 clinicians comprised of 1 occupational therapist, 3 physical therapists, and 2 physicians, 2 clinicians were randomly assigned as the evaluation team for each participant. The team always had 1 certified Assistive Technology Professional with more than 5 years of experience in power wheelchair driving evaluations. The assigned clinicians independently scored every PMRT task during all the driving trials using the following criteria: 4: completed independently, 3: completed hesitantly requiring several trials and minor accidents, 2: commits serious accidents that may cause harm to driver or other people, 1: unable to complete a task. Each clinician had a separate PMRT scoring sheet for each trial.

Repeat Testing:

Subjects underwent all five testing scenarios twice, returning on a second visit in no earlier than 2 weeks.