Zhenghao Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Zhenghao Li has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Mahanth Gowda
Computer Science and Engineering, Pennsylvania State University, University Park, Pennsylvania, USA
5 shared publications - Taiting Lu
Pennsylvania State University, University Park, Pennsylvania, USA
5 shared publications - Yincheng Jin
Computer Science, Binghamton University, Binghamton, New York, USA
4 shared publications - Runze Liu
Pennsylvania State University, University Park, Pennsylvania, USA
3 shared publications - Shengming Yuan
Pennsylvania State University, State College, Pennsylvania, USA
2 shared publications - Xuhai Xu
Columbia University, New York, New York, USA
2 shared publications
Latest Publications
- Open Access
- 2 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-13
iRing: Rich Interaction on Smart Ring with Multi-point Capacitive Touch Screen
- 2 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-13
Open AccessiRing: Rich Interaction on Smart Ring with Multi-point Capacitive Touch Screen
- Open Access
- 30 September 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies
- Vol. 10 (3) , 1-37
AquaTag: An Open-Source Wearable Platform Enabling Flexible Sensor Placement for Aquatic Sensing
Commercial swim wearables increasingly provide performance feedback but remain restricted by fixed, single-placement designs. When worn outside their intended location, the accuracy of motion recognition and swimming analysis drops sharply. Moreover, most existing devices are closed-source, limiting access to raw sensor data and hindering reproducibility and innovation in aquatic HCI research. To address these limitations, we present AquaTag, an open-source aquatic wearable platform that enables flexible placement and consistent motion analysis across different locations. The AquaTag integrates a motion sensor, barometer, and temperature sensor. Using a documented epoxy-potting workflow adapted from established underwater electronics practices, it remains functional during a controlled pressure test equivalent to approximately 14 m water depth for 30 minutes. To support underwater use, AquaTag offers NOR flash or microSD storage for offline logging and Bluetooth for above-water data synchronization. Its power system provides up to 4.4 hours of offline logging or 7.8 hours of non-submerged streaming. The power control and charging mechanisms are designed to meet the waterproofing requirements of the device. The hardware can be manufactured at about $40 per unit. A companion software suite is also developed to annotate and analyze motion data. Specifically, our contributions include: (1) a compact, robust sensing module with a documented, low-cost waterproof packaging workflow suitable for rapid in-house fabrication; and (2) a placement-adaptive machine learning pipeline that delivers accurate swimming analytics from a single motion sensor at flexible body locations. By conditioning a deep learning backbone on the sensor's mounting position, our model learns to infer swimmer kinematics when worn on the head, wrist, or lower back. In leave-one-subject-out evaluation, AquaTag achieved mean macro- F 1 scores of 91.7% for stroke-type classification and 90.6% for swim-state classification across the head, lower-back, and wrist placements; the corresponding best-placement scores were 96.5% and 92.2%, respectively. The AquaTag hardware designs, PCB files, bill of materials, firmware, enclosure CAD files, fabrication instructions, and companion application are released publicly.
- 30 September 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies
- Vol. 10 (3) , 1-37
Open AccessAquaTag: An Open-Source Wearable Platform Enabling Flexible Sensor Placement for Aquatic Sensing
Commercial swim wearables increasingly provide performance feedback but remain restricted by fixed, single-placement designs. When worn outside their intended location, the accuracy of motion recognition and swimming analysis drops sharply. Moreover, most existing devices are closed-source, limiting access to raw sensor data and hindering reproducibility and innovation in aquatic HCI research. To address these limitations, we present AquaTag, an open-source aquatic wearable platform that enables flexible placement and consistent motion analysis across different locations. The AquaTag integrates a motion sensor, barometer, and temperature sensor. Using a documented epoxy-potting workflow adapted from established underwater electronics practices, it remains functional during a controlled pressure test equivalent to approximately 14 m water depth for 30 minutes. To support underwater use, AquaTag offers NOR flash or microSD storage for offline logging and Bluetooth for above-water data synchronization. Its power system provides up to 4.4 hours of offline logging or 7.8 hours of non-submerged streaming. The power control and charging mechanisms are designed to meet the waterproofing requirements of the device. The hardware can be manufactured at about $40 per unit. A companion software suite is also developed to annotate and analyze motion data. Specifically, our contributions include: (1) a compact, robust sensing module with a documented, low-cost waterproof packaging workflow suitable for rapid in-house fabrication; and (2) a placement-adaptive machine learning pipeline that delivers accurate swimming analytics from a single motion sensor at flexible body locations. By conditioning a deep learning backbone on the sensor's mounting position, our model learns to infer swimmer kinematics when worn on the head, wrist, or lower back. In leave-one-subject-out evaluation, AquaTag achieved mean macro- F 1 scores of 91.7% for stroke-type classification and 90.6% for swim-state classification across the head, lower-back, and wrist placements; the corresponding best-placement scores were 96.5% and 92.2%, respectively. The AquaTag hardware designs, PCB files, bill of materials, firmware, enclosure CAD files, fabrication instructions, and companion application are released publicly.
- Open Access
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-24
Toward Scalable ASL Education: Egocentric Stereo Sensing with LLM Feedback for Error-Aware Learning
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-24
Open AccessToward Scalable ASL Education: Egocentric Stereo Sensing with LLM Feedback for Error-Aware Learning
- Open Access
- 2 December 2025
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies
- Vol. 9 (4) , 1-34
FuSenseRing: An Open-Source Platform for Robust Cuffless Blood Pressure Monitoring via Multimodal Sensor Fusion and Temperature-Adaptive Attention
Reliable, frequent blood pressure (BP) monitoring outside clinical settings is crucial for cardiovascular health management, yet current wearable solutions often fail to deliver clinically accurate measurements in real-world conditions. A key limitation stems from the common reliance, particularly in convenient ring form factors, on photoplethysmography (PPG) sensing alone. PPG signals acquired from the finger are highly susceptible to physiological and environmental noise; notably, finger skin temperature fluctuations induce errors exceeding 10 mmHg due to vascular changes, while sensor-skin contact pressure further distorts the signal. This inherent sensitivity of PPG, when uncompensated, prevents many devices from consistently meeting clinical accuracy standards. We introduce FuSenseRing , an open-source smart ring engineered to overcome these issues through synergistic hardware-software co-design. FuSenseRing integrates a multimodal sensor suite (PPG, ECG, skin temperature, contact force) on a custom flexible PCB within a low-cost form factor, approximately $94.9 per unit at retail (batch of 5) and approximately $28.72 per unit at wholesale (batch of 500), while remaining lightweight (4.6 g) to ensure optimal signal acquisition during BP measurements. Our core algorithmic innovation is a Temperature-Adaptive Attention (TAA) mechanism that dynamically fuses PPG/ECG features and recalibrates their contributions using real-time temperature context to specifically mitigate thermal drift. Efficient personalization via transfer learning requires only 5-10 normal-temperature calibration samples. Through rigorous evaluations involving total 60 participants across diverse laboratory-controlled temperatures (17 °C to 40 °C skin temp) and challenging real-world ambulatory conditions (-10 °C to 24.4 °C ambient temp), we demonstrate FuSenseRing's effectiveness. Our system maintains high accuracy across the full temperature range (Overall SBP/DBP ME±STD: -0.09 ± 7.67 / 0.42 ± 5.74 mmHg), satisfying the stringent accuracy criteria of the FDA-recognized AAMI/ISO 81060-2 standard. Furthermore, FuSenseRing significantly outperformed three popular commercial BP rings in the real-world study.
- 2 December 2025
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies
- Vol. 9 (4) , 1-34
Open AccessFuSenseRing: An Open-Source Platform for Robust Cuffless Blood Pressure Monitoring via Multimodal Sensor Fusion and Temperature-Adaptive Attention
Reliable, frequent blood pressure (BP) monitoring outside clinical settings is crucial for cardiovascular health management, yet current wearable solutions often fail to deliver clinically accurate measurements in real-world conditions. A key limitation stems from the common reliance, particularly in convenient ring form factors, on photoplethysmography (PPG) sensing alone. PPG signals acquired from the finger are highly susceptible to physiological and environmental noise; notably, finger skin temperature fluctuations induce errors exceeding 10 mmHg due to vascular changes, while sensor-skin contact pressure further distorts the signal. This inherent sensitivity of PPG, when uncompensated, prevents many devices from consistently meeting clinical accuracy standards. We introduce FuSenseRing , an open-source smart ring engineered to overcome these issues through synergistic hardware-software co-design. FuSenseRing integrates a multimodal sensor suite (PPG, ECG, skin temperature, contact force) on a custom flexible PCB within a low-cost form factor, approximately $94.9 per unit at retail (batch of 5) and approximately $28.72 per unit at wholesale (batch of 500), while remaining lightweight (4.6 g) to ensure optimal signal acquisition during BP measurements. Our core algorithmic innovation is a Temperature-Adaptive Attention (TAA) mechanism that dynamically fuses PPG/ECG features and recalibrates their contributions using real-time temperature context to specifically mitigate thermal drift. Efficient personalization via transfer learning requires only 5-10 normal-temperature calibration samples. Through rigorous evaluations involving total 60 participants across diverse laboratory-controlled temperatures (17 °C to 40 °C skin temp) and challenging real-world ambulatory conditions (-10 °C to 24.4 °C ambient temp), we demonstrate FuSenseRing's effectiveness. Our system maintains high accuracy across the full temperature range (Overall SBP/DBP ME±STD: -0.09 ± 7.67 / 0.42 ± 5.74 mmHg), satisfying the stringent accuracy criteria of the FDA-recognized AAMI/ISO 81060-2 standard. Furthermore, FuSenseRing significantly outperformed three popular commercial BP rings in the real-world study.
- Open Access
- 27 September 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-17
SignGlass: First-Person View Comprehensive and Generalizable ASL Translation Using Wearable Glass
- 27 September 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-17
Open AccessSignGlass: First-Person View Comprehensive and Generalizable ASL Translation Using Wearable Glass