State Key Laboratory of Transducer Technology

We present a wearable intelligent MEMS accelerometer that performs in-sensor action recognition via a physical reservoir formed by a single nonlinear micro-resonant beam. The beam’s hybrid mechanical nonlinearity lifts raw acceleration signals into a high-dimensional state space, enabling rich spatiotemporal feature extraction without pre-processing or calibration circuitry. A lightweight linear read-out, implemented digitally, completes the classification task with negligible latency and power overhead. Prototyped device achieves a 96.44 % accuracy in distinguishing six special human actions. A systematic sensitivity study of reservoir hyperparameters identifies broad performance plateaus and deployment-friendly operating regions across SNRs. Robustness tests that inject shape errors to emulate temperature drift and cross-device dispersion show no performance degradation, indicating resilience to device and environmental variation. Emitting labels, not waveforms, the sensor minimizes data transmission, reducing energy use, improving privacy, and simplifying system integration. The inherent physical reservoir removes the need for adaptive parameter updates, cutting system complexity, energy, and hardware cost while enabling real-time edge operation. These characteristics make the device well suited for health assessment and activity recognition in IoT settings, offering a promising path to efficient on-device intelligence.

ABSTRACT Gas chromatography‐ion mobility spectrometry (GC‐IMS) is a powerful analytical technique for characterizing volatile organic compounds (VOCs). However, variability in retention and drift times frequently complicates peak alignment and downstream analyses. To address this challenge, we introduce RD‐PMA, a Rigid‐Deformable Point‐Matching Algorithm derived from and extending an established point‐set registration framework. RD‐PMA is specifically designed for GC‐IMS data and incorporates dimension‐dependent modeling, where rigid translation is applied in the drift‐time dimension and nonlinear deformation is employed in the retention‐time dimension. Robustness is further enhanced through an iterative outlier‐removal procedure. Extensive evaluations across multiple GC‐IMS datasets, including both homogeneous and heterogeneous conditions, as well as same‐parameter and different‐parameter scenarios, demonstrate that RD‐PMA consistently outperforms conventional alignment approaches, such as internal reference‐based shifting and other point‐matching methods. The algorithm preserves structural consistency while delivering high alignment accuracy and resilience to noise and parameter variability. Collectively, these findings establish RD‐PMA as a reliable and scalable solution for GC‐IMS peak alignment, with broad applicability to large‐scale VOC profiling in chemometrics, food science, and clinical research.

We report for the first time a method for patterning two-dimensional (2D) materials using UV-induced ozone, enabling batch production of high-performance graphene <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{MoS}_{2}$</tex> heterojunction sensors for trace <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{NO}_{2}$</tex> detection. T The proposed methodology achieves precise micronanoscale patterning control while maintaining exceptional material integrity. Through real-time modulation of etching parameters, the process ensures complete protection of underlying layers during sequential patterning steps. The fabricated devices exhibit remarkable sensing performance with a detection limit of 250 ppb and excellent selectivity toward <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{NO}_{2}$</tex>. This method shows significant potential for the mass production of advanced sensors based on 2D materials.

A high-performance piezoelectric micromechanical ultrasonic transducer (PMUT) and an ultra-miniature piezoresistive pressure sensor for arterial signal monitoring have been developed. The radius of PMUT unit is <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$40 \mu ~\mathrm{m}$</tex>, with resonance frequency of 5.5 MHz in air and attenuated to 4.4 MHz in water, exhibiting a -3 dB bandwidth of 59 %. The maximum measured sound pressure is 4.28 kPa/V. The pressure sensor measures <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$0.4 \times 0.4 ~\text{mm}^{2}$</tex>, and its ultraminiature design is conducive to achieving high-precision tactile perception. The pressure sensor demonstrates a fullscale nonlinearity of 0.39 % and an amplified sensitivity of 26.9 mV/kPa. The PMUT and five pressure sensors are mounted on PCB, enabling simultaneous acquisition of radial artery motion waveform and pressure signals. This is the first study combining PMUT and pressure sensors to investigate the mechanical parameters of radial artery. The derived Young's modulus effectively reflects the degree of arterial stiffness, thereby establishing a non-invasive approach for continuous monitoring of arterial stiffness.