Controlled-STM: A two-stage model to predict user’s Perceived Intensity for Multi-point Spatiotemporal Modulation in Ultrasonic Mid-air Haptics

Literature Title

Controlled-STM: A Two-stage Model to Predict User’s Perceived Intensity for Multi-point Spatiotemporal Modulation in Ultrasonic Mid-air Haptics

Literature Information

• Topic Area: Human-Computer Interaction and Ultrasonic Mid-air Haptics Technology
• Keywords: Mid-air haptics, perception, two-stage model, multi-point spatiotemporal modulation, multivariate regression

Research Background and Issues

• What problems or challenges did the authors identify?

  • Multi-point spatiotemporal modulation (STM) in ultrasonic mid-air haptics involves complex parameter combinations (e.g., rendering frequency, number of focal points), which may affect the perceived haptic intensity by users.
  • Current studies overlook the impact of these high-level parameters on physical stimuli, focusing solely on perceived intensity without a comprehensive model of how physical stimuli influence user responses.
  • It remains unclear which physical attributes most significantly affect perceived intensity, hindering effective haptic design.

• Why is this problem important?

  • Quantitative prediction of perceived haptic intensity could help haptic device manufacturers optimize performance and improve design efficiency.
  • Decoupling the relationship between user perception and physical stimulus attributes can provide more scientific guidance for human-computer interaction research.

• Research Motivation and Related Work:

  • Multi-point STM modulation technology can render complex haptic patterns, but systematic predictive models for perceived intensity are lacking.
  • Existing studies attempt to explore the interaction between perception and device parameters, but research on the mechanisms of physical stimuli remains insufficient.

Solution

• What methods or solutions did the authors propose?

  • A two-stage model was proposed:
    1. Physical Model: Predicts physical attributes (e.g., peak sound pressure, peak force) from input parameters.
    2. Perception Model: Predicts user-perceived intensity and inter-user perception differences based on physical attributes.
  • Three experiments were conducted to validate the impact of physical attributes on perceived intensity and establish mathematical models.

• What are the innovative aspects of this solution?

  • For the first time, physical attributes of the device and perceived intensity were separated, with their relationship clarified through theoretical and experimental analysis.
  • A comprehensive predictive framework was introduced, significantly reducing the need for trial-and-error in haptic design.

• What are the implementation steps and key technologies used?

  1. Experimental Design:
     • Measure physical outputs of the device under different parameters (e.g., frequency, number of focal points).
     • Determine the Minimum Perceivable Threshold (MPT) through user studies.
     • Conduct quantitative estimation experiments on user-perceived intensity.
  2. Data Processing and Modeling:
     • Use polynomial regression to fit the physical model.
     • Establish the perception model based on physical attributes and STM parameters.
  3. Validation and Optimization:
     • Validate model accuracy using two test sets, considering various combination conditions for training and validation.

Research Outcomes

• What specific outcomes were achieved?

  • Physical Model:
    • Successfully predicted physical attributes such as peak sound pressure, with an average relative error of 7.8%.
    • Four influencing factors were quantified: nonlinear effects, power limitations, suboptimal driving frequency, and temperature effects.
  • Perception Model:
    • Prediction error for perceived intensity was 8.0%, and prediction error for inter-user perception differences was 8.8%.
    • High correlation between user responses to physical pressure matrices and perceived intensity was confirmed.

• What advantages does it have compared to existing solutions?

  • Enables early adaptation to different device physical performances, predicting user perception through modeling and reducing reliance on device debugging.
  • Accounts for physical performance limitations of devices, enhancing the robustness of perceived intensity predictions.

• What are the experimental or evaluation results?

  • The model demonstrated significantly higher accuracy compared to traditional methods based on linear assumptions of physical parameters.
  • Experiments revealed notable parameter dependencies, such as optimal rendering frequency in the range of 20-40 Hz.

• Limitations and Future Directions

  • Limitations:

    • The model is optimized for specific device types (e.g., OpenMPD, Ultraleap), and its generalizability requires further validation.
    • Limited to specific parameter ranges (e.g., rendering frequency between 10-80 Hz).

  • Future Directions:

    • Explore other variables such as shape rendering and skin contact area effects on perception.
    • Optimize the model to adapt to different ultrasonic haptic devices.
    • Validate the model across broader user groups and more complex scenarios.