RESNA 26th International Annual Confence
Today, there are three competing technologies to provide head-pointing mouse emulation, as exemplified by five products. Since each of these devices claims to provide similar functionality, it can be difficult for the clinician to decide which is the appropriate device for a client. The current study compares the functional performance of the three input technologies using three commercially available devices, the HeadMaster Plus, Tracker 2000, and Tracer. Each device was used to produce a series of drawings of similar complexity until the participants achieved a stable level of performance. The number of trials required to achieve mastery, the speed of drawing at mastery, and the accuracy of drawings was compared for the devices. In addition, the participants were asked about their subjective experience using the devices. Each of the three technologies was fastest for some participants, but the HeadMaster produced the most consistently fast drawing times. Results indicated that, while performance of the devices was similar, participant preference was driven by comfort more than by performance. The two fastest devices, the HeadMaster and Tracer, both resulted in complaints regarding comfort, while the most comfortable device, the Tracker 2000, was preferred by participants even though it was slightly slower in performance.
An individual with a condition such as a high-level spinal cord injury, cerebral palsy, or amyotrophic lateral sclerosis (ALS) may find that, while they are able to control the position of their head easily, they have little or no control over their lower bodies. When computers were controlled entirely by the keyboard, this limitation was a difficulty, but not insuperable constraint on computer access. Technologies such as Morse code, mouth sticks, and head-pointers would allow the generation of keyboard characters (1). However, as the computer interface evolved into the current graphical, mouse-based design, simple discrete selection technologies became cumbersome for controlling the computer. For this interface, some form of mouse-emulation is highly desirable to provide control.
For the individual with good control of head position, but poor control of the lower body, head-pointing technology can provide a very effective replacement for the manual mouse. Kanny and Anson (2) found that a high performance head pointer matched to an accomplished user could provide performance nearly equivalent to the speed of a mouse user. That study compared the performance of the HeadMaster, using ultrasonic technology, to that of the FreeWheel, using infrared reflective technology.
In the decade since that study, new head pointing devices have appeared, and new sensing techniques have also continued to appear. In the current marketplace, there are three technologies for tracking head position: ultrasonic, exemplified by the HeadMaster; infrared reflective, as in the HeadMouse, Tracker 2000, and Smart-Nav; and gyroscopic, as in the Tracer. Each of these technologies has advantages which make it desirable for some populations, but there are no quantitative comparisons of performance of the available technologies to assist the clinician in deciding which product type might be best for an individual who needs a mouse emulator.
This study sought to determine which of the currently available head-pointing technologies provides the fastest and most accurate emulation of the mouse for computer control.
This study used a single-subject, repeated measures design to assess performance of head-pointing technologies. Three commercially available head pointing devices were used in this study: the HeadMaster Plus, from Prentke Romich; the Tracker 2000, from Madentec; and the Tracer, from Boost Technologies. The selection of specific devices within a category was made on the basis of availability, and should not be taken as endorsement of a specific product.
Prior to initiating the study, a series of graphical images, consisting of circles, arcs, and rectangles, were prepared. To assure equal difficulty and complexity, each of the images contained the same elements, simply rearranged on the screen.
The computer workstation was set up with Deneba Canvas, set up with a visible drawing grid and "snap-to-grid" turned on, so that the program assisted in fine precision for the drawings.
A convenience sample of six able-bodied adults (four female and two male) were used in this study. Able bodied subjects were used to separate the effects of disability from the function of the devices being tested. Each subject was asked to reproduce the series of graphic images using a single head-pointer until they produced three drawings with elapsed times within 7% of each other. Once they had achieved proficiency with one device, they were switched to the next device in their series, until they achieved proficiency with that device, at which time they used the third device. After achieving fluency with the third device, the subjects completed a questionnaire about their experience with the three devices, and their preferences between them.
Initially, we assumed that, because the subject would be learning the graphics program and the head-pointer, that the learning of the device should be evaluated only in the second and third devices used by a subject. In fact, the subjects learned to use each head-pointer in between 6 and 8 trials, regardless of position in the sequence. There are no learning advantages in any of these devices.
While not as homogeneous as the number of trials required to achieve mastery, there was a great similarity in times to produce the drawings across subjects as well. For five of the six subjects, the fastest times were between 200 and 220 seconds per drawing at mastery. These minimum times are interesting because they were spread across all three input devices. However, there were clear preferences in the performance of the three devices, with the fastest device for each subject being some 30% faster than the slowest at mastery.
The devices were in a three-way tie for fastest input. Each device proved to be fastest for two subjects. The HeadMaster Plus was second fastest for four subjects and never the slowest. The fastest input at mastery for the Headmaster was 196 seconds, and the slowest 318 seconds. The Tracer, which exhibited a drift problem in our preproduction unit, produced both the fastest and the slowest times at mastery. One subject was able to produce the drawings in 184 seconds at mastery, while another, plagued by drift, required 618 seconds to complete the standard drawings. The fastest user of the Tracker 2000 was able to produce the drawings in 196 seconds, while the slowest required 415 seconds. Both the Tracer and Tracker 2000 required more than the average across devices for half of the subjects, while the HeadMaster required more than the overall average for only one subject.
There did not appear to be any systematic differences in accuracy using these devices. Of the 22 errors in drawings identified across subjects, virtually all were consistent across devices for a given subject, indicating that the errors were caused by the subject's perception of the drawings rather than any specific device's ability to reproduce the drawings.
Although the HeadMaster generally produced faster drawing times, subjects almost universally preferred the Tracker 2000 for its level of comfort. In our study, there were no rewards based on productivity, as there are in a work environment, which may have increased the interest in a device with greater performance.
The results of this study failed to show a distinctly superior technology for head pointing in terms of overall performance. Each of the devices tested proved to be superior for an equal number of subjects, though the HeadMaster was always in the upper region of speed for production. For most subjects, the HeadMaster and Tracer produced times that were very similar, suggesting that these devices are able position the mouse cursor as quickly as a typical human can move his/her head. Further design changes for these two devices should focus on the issue of comfort. For most users, the Tracker was somewhat slower than the other two devices. In this application, which is more demanding than text input, the Tracker 2000 would benefit somewhat from speed enhancements.
In terms of clinical recommendations, this study shows that there is no great performance advantage to any of the devices in question, so the decision regarding which device to recommend can be made in terms of comfort, cost, and availability.