Abstract
Objectives: This dataset collection aims to facilitate research on the transmission of digitized vibrotactile signals over communication networks, emphasizing the importance of minimizing delay and bandwidth requirements while preserving perceptual quality in human sensory experiences. Data description: The dataset includes perceptual ratings derived from two user studies evaluating the quality of vibrotactile signals compressed using various single-point tactile codecs. Participants provided subjective quality ratings for a range of compression ratios to assess data fidelity and the extent to which compression algorithms may compromise quality in favor of file size reduction.
Introduction
As the "Tactile Internet" evolves, the real-time transmission of physical sensations—such as vibrations and textures—becomes a critical requirement. However, raw tactile data is high-dimensional and bandwidth-intensive. To facilitate efficient transmission, signal compression is necessary. The primary challenge lies in balancing file size reduction with human perception; excessive compression can lead to distorted or unnatural sensations.
The VTQA (VibroTactile Quality Assessment) dataset was developed to address the scarcity of standardized, open-access behavioral data. By mapping objective compression ratios to subjective human quality ratings, this work provides a foundation for understanding the "perceptual transparency" of haptic codecs—the threshold at which humans can no longer distinguish between original and compressed signals.
Methods
The dataset aggregates data from two distinct user studies involving 63 participants.
- Experimental Setup: Vibrotactile stimuli were delivered to the pulp of the participant's non-dominant index finger in a controlled environment.
- Task: Participants were presented with pairs of signals consisting of a reference (original) and a test (compressed) signal. They then provided similarity or quality ratings on a numeric scale.
- Codecs Evaluated: The studies tested three state-of-the-art vibrotactile codecs:
- A rate-scalable perceptual wavelet-based codec.
- Pvc-slp: A codec based on sparse linear prediction.
- Vc-pwq: A codec utilizing perceptual wavelet quantization.
- Rigor: To ensure data integrity, the researchers employed counterbalanced stimulus ordering and precise timing for delivery.
Dataset Description
The VTQA dataset is organized according to the Brain Imaging Data Structure (BIDS) standard (Version 1.10.0), adapted for haptic data to ensure interoperability and ease of use for the research community.
The repository includes behavioral data in TSV and JSON formats, participant-level metadata, the original stimuli, and the experimental code required for reproduction. It is openly available on Figshare under a Creative Commons Attribution 4.0 International License.
Key Results and Discussion
As a "Data Note," the primary output is the validated VTQA repository itself. Key insights derived from the data include:
- Compression-Quality Trade-off: The data provides a granular mapping of how subjective quality ratings decline as compression ratios increase, identifying the "sweet spot" for network transmission.
- Algorithmic Benchmarking: The dataset serves as a benchmark for three different algorithmic approaches to tactile signal processing.
- Standardization: Adopting the BIDS format promotes better data-sharing practices within the haptics community, facilitating the integration of behavioral data into machine learning pipelines.
Limitations and Future Work
Users of the dataset should note that it focuses on a specific subset of original signals and compression ratios. Additionally, while 63 participants is a significant sample for haptic research, some specific conditions have smaller subsets due to protocol improvements (e.g., increased sampling rates) made during the study.
The ultimate goal of this data is to serve as a training set for developing automated "objective quality metrics." Such algorithms would predict human ratings without the need for exhaustive user studies, accelerating the development of future haptic codecs.
Summary updated