User Interface Testing of an Assistive Workstation Robot for Persons with High-level Tetraplegia

James J. Wagner, MSME; H.F. Machiel Van der Loos, PhD; J. A. Czaja; V. Punj, MD; B. Jenny Kiratli, PhD

Objectives:The goal of this project is to pilot test the data instruments that were developed for the upcoming user interface testing of the Professional Vocational Assistive Robot (ProVAR). This robot, a workstation-based manipulator to aid the return to vocational activity of persons with high-level SCI tetraplegia, will undergo significant development and testing in the coming 3 years. ProVAR has a novel interface designed to facilitate the learning of its functions and the operation of its features. The data instruments are geared to understanding how users react to the system and gain personal benefit in terms of quality of life and functional independence.

Methods: Six subjects, two with high-level tetraplegia and four who are able-bodied, learned to operate the robot's interface. Subjects with tetraplegia were offered use of a head motion cursor control device, a trackball, and voice recognition software. Able-bodied subjects were given a standard mouse and keyboard. Subjects were trained on the robot for 20 minutes by the same instructor. They were familiarized with the computer, the 2-window browser environment, and the robot interface itself. They were given 90 minutes to program the task of having the robot insert a videotape into a player. A 65-question survey was then administered covering the categories indicated in Figure 1. A score of less than 4 indicates that the feature "needs improvement." In addition, individual programming steps were logged by the interface computer to calculate time-to-completion and learning effects.

Results: Subjects reported overall satisfaction with the interface in all the categories except "error messages" (Figure 1). When considered individually, only 9 of the 65 questions rated below 4 on average, revealing that ease of use and feedback were the primary areas of needed improvement. There was a discrepancy in certain areas between able-bodied and tetraplegic subjects. Notably, people with tetraplegia were more patient and became less frustrated, but they found the interface more complex than did able-bodied subjects. The average amount of time to program each step decreased from 120 seconds to 90 seconds between the beginning and end of the session, likely indicating a learning effect (no statistical analysis was done due to the small sample size in this pilot study).

Conclusions: These assessment and logging tools, adapted from more general computer user interface literature, were shown to be effective in capturing the various aspects of the interaction between a user and a workstation robot. Outcomes of this initial interface testing will be used to implement improvements in the interface as well as the evaluation process before extensive evaluation of ProVAR begins with veterans who have high-level tetraplegia.

Funding: The pilot study was funded by the Paralyzed Veterans of America Summer Scholars program. The ProVAR project is funded by VA Rehabilitation R&D Service grant A2684I.