In-vehicle Performance and Distraction for Midair and Touch Directional Gestures

Title of the Paper

In-vehicle Performance and Distraction for Midair and Touch Directional Gestures

Paper Information

• Research Domain: Human-Computer Interaction, In-vehicle Interface Design
• Keywords: In-vehicle interaction, gesture recognition, touchscreen, midair gestures, driving tasks, attention distraction

Research Background and Issues

• Problems and Challenges:

  1. In-vehicle systems are increasingly relying on touchscreens as the primary input device, which requires drivers to divert their gaze from the road, leading to attention distraction issues.
  2. Although midair gestures have been adopted by some high-end vehicles (e.g., BMW 7 Series, Mercedes-Benz S-Class), there is limited research comparing their performance with traditional touchscreen operations, especially for short-term selection tasks involving multi-directional gestures.

• Research Significance: Input method performance and potential distraction in driving environments are critical for driving safety. Developing safer and more efficient in-vehicle interaction methods can enhance the driving experience and reduce accident risks.

• Motivation and Related Work:

  • Existing studies suggest that midair gestures can reduce visual attention shifts, but their reliability and usability are lower compared to touchscreen interactions.
  • Furthermore, while directional gestures dominate user preference surveys, systematic and controlled studies on their performance remain scarce.

Solution

• Research Methods or Solutions: This study directly compares the performance and distraction levels of the following three in-vehicle input methods:

  1. Midair gestures (8 directions, triggered by pinch gestures using the index finger and thumb).
  2. Touchscreen swiping (8 directions).
  3. Touchscreen tapping (8 buttons, used as a baseline evaluation).

• Innovations:

  1. Introduced the "atomic input task" approach, focusing on single atomic operations to ensure high internal validity.
  2. Expanded input tasks to 8 directions, surpassing previous studies that only included 1-2 directions.
  3. Evaluated distraction during driving tasks using the Lane Change Task (LCT) and Standard Deviation of Lane Position (SDLP) metrics.

• Implementation Steps and Key Techniques:

  1. Experimental setup: Utilized a driving simulator, Logitech G29 steering wheel, Android tablet, Vicon motion tracking system, and cameras to record drivers' gaze behavior.
  2. Experimental design: Participants simultaneously performed the Lane Change Task and input tasks, with measurements of reaction time, completion time, accuracy, and distraction metrics (e.g., number of glances away from the screen).
  3. Data analysis: Examined differences in selection time, accuracy, SDLP, driving speed, and glances across the three input methods.

Research Findings

• Specific Findings:

  1. Performance Comparison:
     • Midair gestures had faster selection times than touch-based methods but slightly lower accuracy.
     • Touchscreen swiping demonstrated higher accuracy but slower selection times.
     • Touchscreen tapping achieved the best accuracy but required the most visual attention shifts.
  2. Attention Distraction:
     • Midair gestures significantly reduced the number of glances away from the screen, causing the least interference with driving tasks.
     • Touchscreen tapping caused the most distraction, with swiping falling in between.
  3. Participant Diversity:
     • Midair gestures exhibited greater individual differences. Some participants completed tasks quickly and accurately, while others showed higher error rates or slower reaction times.

• Advantages Compared to Existing Solutions:

  • This study is the first to directly compare 8-direction midair gestures with touchscreen inputs in a highly controlled experiment.
  • Provides strong evidence supporting midair gestures in reducing driving distraction and designs examples applicable to various driving scenarios.

• Experimental or Evaluation Results:

  • Midair gestures: Fast selection time (1004 ms), moderate accuracy (82%), lowest distraction (average 3.7 glances away from the screen).
  • Touchscreen swiping: Moderate selection time (1296 ms), high accuracy (95%), moderate distraction.
  • Touchscreen tapping: Slow selection time (1150 ms), highest accuracy (98%), highest distraction (10 glances away from the screen).

• Limitations and Future Directions:

  1. Limitations:
     • Conducted in a laboratory driving simulator, lacking the complexity of real-world driving scenarios.
     • Did not incorporate ultrasonic haptic feedback to further enhance midair gesture performance.
  2. Future Research Directions:
     • Validate findings in high-fidelity driving simulators or real-world driving environments.
     • Explore personalized midair gesture recognition parameters using machine learning.
     • Investigate midair gesture interaction with ultrasonic haptic feedback.
     • Study individual differences and learning curves to understand their impact on input performance.