STTR Phase I: Compact, Comfortable Wristband for High-Fidelity Gesture Recognition via Distributed Strain Sensing, Analog Multiplexing, and Edge Machine Learning

Award ID: 2538142

The broader/commercial impact of this Small Business Technology Transfer (STTR) Phase I project is to overcome a longstanding limitation in how people interact with computers. Most digital work today relies on input tools originally designed for flat screens, even as computing increasingly moves into three-dimensional environments used for design, analysis, communication, and accessibility. This mismatch affects tens of millions of Americans who use digital systems for work, education, and creative activities, leading to inefficiency, physical discomfort, and barriers to participation. This project addresses the problem by developing a wrist-worn interface that allows users to interact with three-dimensional software through natural hand movement rather than rigid controllers or keyboards. The innovation advances scientific and technological understanding by demonstrating how wearable sensing and data-driven interpretation of human motion can enable intuitive interaction across diverse real-world settings. The technology is situated within the field of human-computer interaction, with an initial market focus on professional design and engineering workflows where productivity, comfort, and seamless integration are critical. The value proposition is driven by a proprietary approach that tightly couples wearable sensing with software-level interpretation, creating a durable competitive advantage. By year three, the technology is projected to reach early professional users nationwide, with impact measured through adoption, workflow efficiency, and reduced physical strain. This Small Business Technology Transfer (STTR) Phase I project investigates a wearable sensing and decoding architecture for enabling continuous and discrete hand interaction using wrist-mounted strain sensing. The technical problem addressed is whether soft, body-worn sensors can reliably infer hand intent with sufficient bandwidth and stability to support real-time interaction, while avoiding the power consumption, occlusion, and form-factor limitations of vision-based or rigid input systems. The research objective is to evaluate whether distributed strain sensing coupled with compact signal processing and learning-based inference can support robust hand state estimation under realistic use conditions. The proposed research integrates compliant textile-based sensing elements with an ultra-low-power analog front end that encodes multiple sensing channels onto a shared signal pathway. Timefrequency features derived from the encoded signal are processed using a lightweight machine learning architecture that estimates reduced-order continuous hand state alongside discrete interaction events. Phase I evaluation emphasizes signal separability, decoding accuracy, latency, and robustness, assessed through benchtop characterization and controlled user studies using independent motion reference data. Anticipated results include evidence that wrist-mounted strain sensing with analog spectral encoding can support reliable, low-power hand interaction, establishing a scalable technical foundation for subsequent system refinement and broader interactive applications. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Program Director: Lindsay Portnoy