Assessing the Usability of Gaze-Adapted Interface against - - PowerPoint PPT Presentation

Institute for Web Science and Technologies · University of Koblenz-Landau, Germany

Assessing the Usability of Gaze-Adapted Interface against Conventional Eye-based Input Emulation

Chandan Kumar, Raphael Menges and Steffen Staab

CBMS17 ST7 - MULTIMODAL INTERFACES FOR NATURAL HUMAN COMPUTER INTERACTION: THEORY AND APPLICATIONS

Motivation

2

Assessing the Usability of Gaze-Adapted Interface against Conventional Eye-based Input Emulation

• Social platforms are an opportunity for physically

impaired people to connect with others

• Eye gaze tracking is an emerging input device
• Two interface approaches to include eye gaze

– Emulation of traditional input devices – Gaze-adapted interface

Research Question: What is the impact on Usability and Task Load for the user?

Assessment of Usability

3

Assessing the Usability of Gaze-Adapted Interface against Conventional Eye-based Input Emulation

Emulation1 Gaze-Adapted Twitter

1OptiKey Software and

Firefox, showing mobile Twitter page

There are two major challenges1 for eye tracking

Eye Tracking

Assessing the Usability of Gaze-Adapted Interface against Conventional Eye-based Input Emulation

• Limited Accuracy
• Maximal accuracy is one degree

due eye geometry

• Calibration drift through head

movements → Size and position of interface elements

• Midas Touch
• Eye is both sensor and controller

→ Dwell time based interaction

Idle Fixation Gaze Fixation

1Kumar, C., Menges, R., & Staab, S. (2016). Eye-Controlled

Interfaces for Multimedia Interaction. IEEE Multimedia, 23(4), 6-13. 4

Emulation of traditional input devices

5

Assessing the Usability of Gaze-Adapted Interface against Conventional Eye-based Input Emulation

• Emulation of mouse and keyboard using gaze
• Dwell time based button interaction
• Example of left mouse button click

Dwell on left mouse click button → Dwell on click area → Magnification of area and another dwell on exact position

Example application: https://github.com/OptiKey/OptiKey

Gaze-adapted Twitter

6

Assessing the Usability of Gaze-Adapted Interface against Conventional Eye-based Input Emulation

• Content Area displays recent tweets and provides no interaction
• Action Bar provides contextual actions by dwell time buttons

Gaze-adapted Twitter: Demo

7

Assessing the Usability of Gaze-Adapted Interface against Conventional Eye-based Input Emulation

• Hardware

– Tobii EyeX consumer eye tracking device

• Software

– Our gaze-adapted Twitter application – OptiKey operating Firefox with mobile Twitter page

• Study

– Learning: Eye tracking tutorial provided by Tobii executed – Think-aloud study, including SUS and NASA-TLX survey – Counter-balancing between the two softwares performed

Experimental Setup

8

Assessing the Usability of Gaze-Adapted Interface against Conventional Eye-based Input Emulation

• Task

– Write a tweet and publish it, find a particular user and follow her, find and like a certain tweet. Explore the application (5-10 min)

• Participants

– 13 students (10M, 3F), aged between 20 and 39

Results: System Usability Score

9

Assessing the Usability of Gaze-Adapted Interface against Conventional Eye-based Input Emulation

P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 100 80 60 40 20

Gaze-adapted T witter OptiKey (Mobile T witter)

Better

SUS Average1 = 68

Mean = 72 Mean = 50 p =.0044 < 5%

1https://measuringu.com/sus

Results: Task Load Average Raw Score

10

Assessing the Usability of Gaze-Adapted Interface against Conventional Eye-based Input Emulation

Mental Demand Physical Temporal Demand Demand Perfor- mance Effort Frustration 60 40 20 80 100

Gaze-adapted Twitter OptiKey (Mobile Twitter)

Better

Mean = 47 p =.0238 < 5% Mean = 64

Mental Demand Physical Temporal Demand Demand Perfor- mance Effort Frustration 0.00 0.20 0.15 0.10 0.05 0.25

Gaze-adapted Twitter OptiKey (Mobile Twitter)

Results: Task Load Average Weightings

11

Assessing the Usability of Gaze-Adapted Interface against Conventional Eye-based Input Emulation

• Frustration weighting being two times higher for the emulation
• Mental demand, effort and frustration were judged as the most

relevant scales by the participants

Providing the importance

• f the values

Mental Demand Physical Temporal Demand Demand Perfor- mance Effort Frustration 60 40 20 80 100

Gaze-adapted Twitter OptiKey (Mobile Twitter)

Results: Task Load Average Raw Score

12

Assessing the Usability of Gaze-Adapted Interface against Conventional Eye-based Input Emulation

Pairwise weighting indicates importance! Better

• Participants felt stressed when the interface

reacted constantly to their gaze

• Users are very focused on the visual search task

and overlook system’s help (e.g., auto text suggestions while typing)

• Participants prefer the option to personalize the

interaction with respect to their experience

Observations

13

Assessing the Usability of Gaze-Adapted Interface against Conventional Eye-based Input Emulation

Institute for Web Science and Technologies · University of Koblenz-Landau, Germany

Thank you for your Attention!

Chandan Kumar, Raphael Menges and Steffen Staab

Conclusion

• Gaze-adapted interface for Twitter was presented
• Evaluation showed an advantage in both usability and mental

demand for gaze-adapted interface over emulation approach

• Future Work: We implement and evaluate Web browsing with

eyes and mind, for gaze-adaption of various service interfaces

This work is part of project MAMEM that has received funding from the European Union's Horizon 2020 research and innovation program under grant agreement number: 644780.