1. Completion Time Predictions for Secondary Tasks in Non-Stationary Environments
Sc.D. Dissertation Proposal Presentation
Martin J. Schedlbauer
April 28, 2005
Chair:
Committee:
Dr. Jesse M. Heines, University of Massachusetts Lowell
Dr. Giampiero Pecelli, University of Massachusetts Lowell
Dr. Holly Yanco, University of Massachusetts Lowell
Dr. Robert Pastel, Michigan Technological University

2. Motivation
Mobile computing devices are becoming increasingly prevalent:
PDA
Mobile Phones
Personal GPS
Navigation systems:
In-Vehicle Information Systems
Marine Navigation Systems

3. The Problem
Current human-computer interface design is geared toward desktop computing.
Interaction patterns and task completion times are likely affected when the user interacts with an application in a non-stationary environment.
The interaction with the application is frequently a secondary task.

4. An Example Environment

5. Example: Navigation System
Maptech Navigator
12” LCD display
Touch screen
Electronic charting
GPS navigation
Operational data

6. Task Completion Time
Task Completion Time (TT) is defined as the time it takes to perform a series of sub-tasks that constitute an interactive transaction with a system:
Completion time increases in dual-task situations.
where LT is learning time, RT is the reaction (or recognition) time, MT is the mean movement time, and GT is glance time

7. HCI Engineering Models
Movement Time Predictions
Fitts’ Law and recent variations, including Meyer’s Law, Kvålseth’s Law, Oel’s Formulation
Reaction Time Predictions
Hick-Hyman Law
Kvålseth's Decision Model
Learning Time Predictions
Power Law of Practice
Heathcote’s Exponential Law of Practice

8. Fitts’ Law
where a and k are empirically derived coefficients, D is distance to target (amplitude of movement), and We is effective width of the target.
Derived from Shannon-Hartley Theorem
Different formulations by Fitts, Welford, and MacKenzie determine value of ε
Describes univariate pointing tasks
Shown to apply to many different input devices

9. Bivariate Pointing Tasks
Target W D H W' φ
Movement along two dimensions
Cannot simply use width of target
Angle of approach has effect on T
MacKenzie-Buxton
Accot-Zhai

10. Alternative Models
Some alternative models provide more accurate predictions in many cases:
Meyer’s Law
Kvålseth's Law

11. Hick­Hyman Law
Model for predicting reaction time when making a decision among n choices:
Alternative model by Kvålseth:
where k is empirically derived, pi is the probability of choice i.

12. Law of Practice
The time it takes to perform a task on the nth try can be stated as:
Alternative model by Heathcote et al.:
where a is an empirically derived constant and T1 is the time of first try.

13. Relevant Research Applicable research concerning:
small targets
moving targets
non-rectangular targets
expanding targets
area cursors
goal crossing tasks
steering tasks
Touch input and probe width corrections
Additionally, throughput varies with input device, lag and gain
ISO9241-9

14. Input Methods & Devices
Since the research will be focused on marine and automotive navigation systems, a survey of present systems in use resulted in the following findings:
Touch screens are employed more frequently
Trackball, isometric joystick, cursor rocker pads, and fixed buttons are most common
Variable footprint devices are not in use

15. Other Relevant Research
Besides Fitts’ Law, other related research is of importance:
In-Vehicle Navigation Systems
Dual-Task Situations
15-Second Rule
Psychology
Task Scheduling Patterns
Kinesiology
Movement theories

16. Kinesthetic Models Different theories of kinesiology:
iterative corrections model
series of discrete submovements
impulse variability model
initial muscle impulse then gliding
optimized initial impulse model
initial impulse, some corrective submovements
Models presume a closed-loop with visual and proprioceptive feedback.

17. Research Questions
Does Fitts’ Law hold in non-stationary environments?
Does movement time increase during secondary tasks?
How can task completion time be calculated?
How should user interfaces for mobile computing systems be designed?

18. Hypotheses
H10: For rapid aiming tasks in non-stationary environments, the correlation between MT and ID is as predicted by the General Fitts formulation.
H20: Posture does not have an effect on MT in a rapid aiming task when using either finger or stylus as a direct input probe.

19. Hypotheses (cont.)
H30: Performing rapid aiming tasks in a dual-task situation does not have an effect on MT.
H40: Acquisition of vibrating targets can be predicted by the General Fitts formulation.

20. Hypotheses (cont.)
H50: There is no difference in ID and error rate for target acquisition using a trackball compared to using a finger/stylus touch in a non-stationary environment compared to a stationary one.
H6a: TT in non-stationary and stationary environments are positively correlated.
H7a: TT for secondary tasks in a non-stationary and stationary environments are positively correlated.
H8a: TT is proportional to MT + GT.

21. Experiments
The hypotheses will be investigated experimentally through a series of laboratory and field tests with human subjects.
The experiments will be carried out using a specially developed Java software platform.

22. Experiment I Control experiment with general Fitts tasks:
Randomly placed square targets of different sizes
Acquisition with finger and stylus, trackball, and isometric joystick
Sitting and standing posture
20-25 participants

23. Experiment II
Acquisition of vibrating targets that simulate motion in the environment:
Touch screen with finger and stylus only
Several variations of the frequency and amplitude of vibration
Variation of target distance and location, and possibly extent
Standing versus sitting posture
Same participants

24. Experiment III Performance of Fitts tasks is a dual-task situation:
Setup as in experiment I
Monitoring of a “depth indicator” and oral statement when threshold is reached
Measures glance time and effect of cognitive tasks on motor skills
20-25 participants

25. Experiment IV
If time allows, subjects carry out Fitts tasks while walking:
Two speeds of walking
Randomly placed square targets of varying sizes
Touch input only with finger and stylus
10-15 participants that are a subset of the subjects of experiment I

26. Experiment V
Fitts tasks and data entry in a non-stationary environment:
Single and dual-task situations
Carried out on an actual boat
Accelerometer and pendulum apparatus used to quantify motion
8-12 participants

27. Summary of Research Goals
Investigation of the applicability of Fitts’ Law and other movement time prediction models to
non-stationary environments with standing posture
secondary tasks in dual-task situations
Development of an empirically derived predictive model for completion time of secondary tasks in non-stationary environments.
Heuristics for user interface design of mobile computing applications with a specific focus on marine navigation systems.

28. Dissertation Schedule

29. Questions & Answers ?
Copies of the proposal are available for those interested.
Research results will be posted at:
Comments and feedback are welcome.

30. SCINS Survey

31. Typical SCINS
Numeric keypad for waypoint entry
Soft buttons whose purpose depends on context
Rocker-type cursor pad that moves cursor up, down, left, and right

32. Usability Model

33. Moving Targets
Jagacinski et al.
Hoffmann

34. Dual-Task Situations
In situations, where users perform two tasks (primary and secondary), completion time is affected:
Learning time and reaction time can be made constant.