RESNA 27th International Annual Confence

Technology & Disability: Research, Design, Practice & Policy

June 18 to June 22, 2004
Orlando, Florida


Evaluation of Mouse Emulation Using the Wheelchair Joystick

Edmund F. LoPresti1,2, Barry A. Romich1,3, Katya J. Hill4 ; Donald M. Spaeth1
1University of Pittsburgh, Pittsburgh, PA 15260;
2AT Sciences, Pittsburgh, PA 15213;
3
Prentke Romich Company, Wooster, OH 44691;
4
Edinboro University of Pennsylvania, Edinboro, PA 16412

ABSTRACT

A Joystick to Mouse Adapter (JMA) has been designed which allows a power wheelchair joystick to provide proportional control of a computer or augmentative and alternative communication (AAC) system. The JMA has three modes: proportional pointing, proportional directed scanning, and a hybrid of the two. Fifteen wheelchair users performed computer exercises using the JMA as well as a system which converted the joystick signal into four switches. Subjects tended to have faster performance with proportional pointing for an icon selection task and proportional directed scanning for a text entry task, and higher accuracy with proportional directed scanning for both tasks.

KEYWORDS

Integrated controls, power wheelchair, computer access, AAC, performance measurement

BACKGROUND

Many people who use power wheelchairs also require access to computers and other assistive technologies, such as Augmentative and Alternative Vommunication (AAC) devices and Electronic Aids to Daily Living (EADLs). Often, each device has a separate input method. However, a person may have a limited ability to physically operate these devices, and may only be able to achieve reliable and effective control at a single site (1). In this situation, the person may attempt to control multiple input devices mounted at the same site, or may rely on a caregiver to switch input devices when the person wishes to change tasks. A third alternative is to use an integrated control system, which allows a person to operate several pieces of assistive equipment through a single, universal input device (2).

RESEARCH QUESTION

The goal of this study was to evaluate four ways to control a computer cursor using a powered wheelchair joystick. Use of a powered wheelchair joystick to emulate the computer mouse would allow someone to use a single interface – the joystick – to operate his or her wheelchair, computer, and/or a high-tech AAC system.

The first objective of this study was to determine whether a system offering proportional control of the computer cursor would provide improved accuracy and speed compared to a system which converts the joystick into four switches. The second objective was to determine which of three proportional control methods provided the highest accuracy and speed.

METHOD

A Joystick to Mouse Adapter (JMA) was developed which provides three modes of control: proportional pointing (position mode), proportional directed scanning (velocity mode), and a hybrid of the two (3). These conditions may be summarized as follows:

  1. Position Mode: the joystick position determines the position of the cursor on the screen; if the joystick is moved to a certain angle, the cursor moves to a certain location on the screen.
  2. Velocity Mode: the joystick position determines the speed and direction of cursor movement; if the joystick is held at a certain angle, the cursor moves in the indicated direction at a speed determined by the joystick angle.
  3. Hybrid Mode: Position control is active in a region at the center of the joystick range, and velocity control is active outside this region.

The Adaptive Switch Laboratories (ASL) Mouse Emulator provides switch-controlled two-speed on-axis and diagonal directed scanning. Both the JMA and the ASL Mouse Emulator were used to control the cursor on a Windows computer through an Invacare Mk. IV power wheelchair joystick (Invacare, Elyria, OH).

Fifteen subjects took part in this study (20-65 years of age, mean 39.7 years). Subjects were currently using a powered wheelchair with a hand-operated joystick. Eight subjects had cerebral palsy, three had spinal cord injuries, one had multiple sclerosis, one had muscular dystrophy, and two did not report their diagnoses.

Subjects were asked to perform two tasks using each of the four modes of control. The first task was an icon selection exercise. At the beginning of the exercise, a black circle appeared at the center of the screen. The subject held the cursor within this circle, and a 15 mm target symbol then appeared elsewhere on the computer screen. The subject would select the target by “dwelling” on the target for 500 ms, then return the cursor to the center of the screen. Targets appeared at any of four distances from the center of the screen. Targets were presented along the vertical and horizontal axes, and at randomly-selected off-axis locations. For each condition, the subject was asked to perform four repetitions of the exercise, with 28 targets presented in a random order in each repetition. The first repetition was treated as practice, and data from the remaining three repetitions were used for analysis. Due to time limitations, nine subjects were unable to complete four repetitions for each condition. In these cases, data from the same number of trials were used for data analysis in each condition for that subject. For example, if a subject only completed three repetitions in one condition, data from the second and third repetitions were used for each condition.

The second task was the transcription of text using onscreen keyboard software (WiViK Version 2.5, Prentke Romich Company, Wooster, OH) and Typing Instructor Version 11 (Individual Software, Pleasanton, CA). For each condition, the subject entered four sentences or as much text as possible within five minutes. Letters were entered using the onscreen keyboard by moving the cursor to the desired letter and pressing a switch which acted as the mouse button.

Table 1: Icon Selection Performance (Mean ± Standard Deviation)

 

Throughput (bits/sec)

Accuracy

Position Mode

0.94 ± 0.58

69% ± 33%

Velocity Mode

0.65 ± 0.33

86% ± 30%

Hybrid Mode

0.74 ± 0.43

78% ± 26%

Switched Joystick

0.56 ± 0.20

84% ± 27%

RESULTS

Results for the icon selection task are shown in Table 1 and Figure 1. Throughput is a measure of speed which accounts for the size of the targets and the distance traveled to reach the target (4). Accuracy is the percentage of targets successfully acquired. Throughput was significantly higher in the position mode than any other condition ( a < 0.05), whereas accuracy was significantly higher for the velocity mode compared to the position mode ( a < 0.05).

Table 2: Text Entry Performance (Mean ± Standard Deviation)

 

Speed (words/min)

Accuracy

Position Mode

1.7 ± 1.7

54% ± 45%

Velocity Mode

2.0 ± 1.4

83% ± 34%

Hybrid Mode

1.8 ± 2.0

59% ± 44%

Switched Joystick

1.4 ± 1.1

80% ± 34%

For the text entry task, speed was measured in words per minute. Accuracy was measured as the number of characters entered correctly divided by the number of characters typed, presented as percentage of characters. Subjects were not able to make corrections. Results for this exercise are shown in Table 2 and Figure 2. Speed was significantly higher in the velocity mode compared to the position mode or switched joystick condition ( a < 0.1), and accuracy was significantly higher for both the velocity mode and switched joystick condition compared to both the position and hybrid modes ( a < 0.05).

Figure 1. Icon Selection Performance (Click image for larger view) d

DISCUSSION

For the icon selection task, the data indicate a trade-off between speed and accuracy, in which the position and hybrid modes offer increased speed of performance at the cost of accuracy. The velocity mode tends to offer the best performance in general, having the highest accuracy and increased speed compared to the switched joystick for both tasks, and also the highest speed for the text entry task. However, for clients who are able to use the position or hybrid modes accurately, these modes can allow faster performance. Subjects who were able to use the position and hybrid modes to achieve icon selection accuracy comparable to accuracy with the switch mode (accuracy in pointing mode at least 95% of accuracy in switched mode) still had significantly higher throughput with the pointing mode compared to velocity and switched modes ( a < 0.01).

Figure 2: Text Entry Performance (Click image for larger view) d

Performance may also depend on whether the client needs to use the mouse button. Three subjects were able to use the position and hybrid modes (with varying degrees of success) for the icon selection task, but not for the text entry task. This was due in part to the fact that subjects had to use the mouse button for the text entry task, but not the icon selection task. In the velocity mode and switched joystick conditions, subjects could take their hand away from the joystick and press a switch, and the cursor would remain in place. In the pointing and hybrid modes, the cursor would tend to return to the center of the screen if the joystick was released. Therefore, subjects needed two hands to perform the text entry task, one to control the joystick and one to control the switch. Since subjects with limited movement are among the main candidates for integrated control systems, software that removes the need for a mouse button (such as software that supports a dwell mode) may be desirable in general.

REFERENCES

  1. Angelo J, & Trefler E. (1998). A Survey of Persons Who use Integrated Control Devices. Assistive Technology. 10 :77-83
  2. Guerette P, & Sumi E. (1994). Integrating Control of Multiple Assistive Devices: A Retrospective Review. Assistive Technology. 6 :67-76.
  3. Romich, B.A., LoPresti, E.F., Hill, K.J., Spaeth, D.M., Young, N.A., Springsteen, J.P. (2002). Mouse Emulation Using the Wheelchair Joystick: Preliminary Performance Comparison Using Four Modes of Control. Proceedings of the RESNA 2002 Annual Conference. pp. 106-108.
  4. Douglas, A.S., Kirkpatrick, A.E., MacKenzie, I.S. Testing Pointing Device Performance and User Assessment with the ISO 9241, Part 9 Standard. Proc. CHI . pp. 215-222.

ACKNOWLEDGMENTS

Funding for this study was provided by the University of Pittsburgh Rehabilitation Engineering Research Center on Wheeled Mobility, National Institute on Disability and Rehabilitation Research, US Department of Education, Washington, DC (Grant #H133E990001).

Author Contact Information:

Edmund LoPresti, Ph.D.,
AT Sciences,
160 N. Craig St. Suite 117,
Pittsburgh, PA 15213,
Phone: 412-687-1181,
Fax: 412-383-6597,
E-mail: edlopresti@acm.org

RESNA Conference Logo