Flexel: A Modular Floor Interface for Room-Scale Tactile Sensing

Document Title

Flexel: A Modular Floor Interface for Room-Scale Tactile Sensing

Document Information

• Subject Area: Human-Computer Interaction (HCI), Tactile Sensing, Smart Indoor Environments
• Keywords: Floor Interface, Architectural Design, Tactile Sensing, Interaction Design, Footprint Tracking, Gesture Recognition, Object Localization, Information Visualization

Research Background and Problem Statement

• Problems and Challenges:

  • Current floor systems lack sensing intelligence and cannot capture vibration and load signals from daily activities, limiting their potential as sensing interfaces.
  • Indoor tactile sensing requires high spatial resolution sensor layouts, but increasing the number of sensors significantly raises costs and system complexity.
  • The design of floor interfaces must adhere to building codes while ensuring maintainability, low cost, and multifunctionality.

• Significance:

  • Floor-based sensing methods are less intrusive, burden-free, and more privacy-preserving compared to visual sensors or wearable devices.
  • Optimizing the design to make floor interfaces more feasible and widely applicable can provide novel solutions for fields such as smart homes and healthcare.

• Research Motivation and Related Work:

  • Previous research has primarily focused on perception systems for specific tasks but lacks broadly applicable floor design guidelines.
  • A systematic analysis is needed to determine suitable hardware architectures for floor interfaces and how to balance sensor density with cost and performance.

Solution

• Method and Solution:

  • Proposed a modular floor interface (Flexel) that achieves room-scale activity sensing through a sparse layout of tactile sensors.
  • Flexel integrates best practices from architectural design and tactile sensing, supporting multifunctional applications (e.g., footprint tracking, foot gesture recognition, and object localization).
  • The system includes modular sensing hardware design, a real-time data visualization graphical user interface (GUI), multiple input analysis functions, and 3D tactile inference capabilities.

• Innovations:

  • Established design guidelines for floor interfaces, including sensor density optimization, modular hardware design, and architecture selection.
  • Developed a structured floor model that achieves high-precision activity sensing through sparse sensing (sensor density 1/81 of traditional pressure mats).
  • Combined user foot data with 3D position inference to enable tactile activity recognition across floor and furniture areas.

• Implementation Steps and Key Technologies:

  1. Hardware Module Design: Each floor module contains multiple load sensors, supporting measurements of pressure intensity and center of weight.
  2. Sensor Density Optimization: Background experiments identified optimal sensor spacing (e.g., 16 cm for footprint center detection, 18 cm for foot direction detection).
  3. System Architecture: Includes hardware module connections, real-time data aggregation into a unified "floor image" format, and subsequent analysis modules.
  4. Prototype Development and Evaluation: Built 50 hardware modules covering a 5 m² area, testing the effects of surface materials, weight measurement accuracy, and other performance metrics.

Research Outcomes

• Specific Results:

  • Proposed a sparse sensing layout that balances system performance and economic cost.
  • Enabled multiple room-scale activity sensing applications: such as footprint tracking, foot gesture recognition, and object localization.
  • Successfully validated the ability to infer desktop operation positions based on "user area weight shifts."

• Advantages Comparison:

  • Compared to traditional pressure mats: Sensor count reduced to 1/81, significantly lowering costs while maintaining functional performance.
  • Compared to other indoor sensing technologies (e.g., vision or RF sensing): Offers better privacy protection and burden-free deployment.

• Experimental or Evaluation Results:

  1. Weight Measurement Accuracy: Central units had a weight error margin of ±12 grams; edge units showed slightly higher deviations.
  2. Surface Material Impact: Floor covering materials had some effect on sensing performance but did not significantly reduce resolution within reasonable ranges.
  3. 3D Tactile Inference: Using an SVM machine learning model, inferred contact position information based on user area weight shifts on the floor, achieving an accuracy of approximately 51.5%.

• Limitations and Future Directions:

  • Limitations:
    • Current system costs remain high, unsuitable for large-scale societal deployment.
    • Only supports vertical force sensing, lacking detection of lateral or shear forces.
    • Limited communication bandwidth requires protocol optimization (e.g., adopting Ethernet).
  • Future Directions:
    1. Enhance load sensor accuracy to support more degrees of freedom (e.g., shear sensing).
    2. Expand application scenarios, such as health monitoring and behavior recognition.
    3. Investigate more user and scenario data to improve system generalizability.

Conclusion

• Flexel provides an innovative and practical solution for indoor user and environment sensing. Through modular design and optimized sparse sensor layouts, it demonstrates best practices for balancing cost-effectiveness and performance. This achievement paves the way for future smart home and indoor space designs, showcasing the innovative potential in the field of human-computer interaction.