Pedestrian Pose Recognition Based on Frequency-Modulated Continuous-Wave Radar with Meta-Learning.

With the continuous advancement of autonomous driving and monitoring technologies, there is increasing attention on non-intrusive target monitoring and recognition. This paper proposes an ArcFace SE-attention model-agnostic meta-learning approach (AS-MAML) by integrating attention mechanisms into residual networks for pedestrian gait recognition using frequency-modulated continuous-wave (FMCW) millimeter-wave radar through meta-learning. We enhance the feature extraction capability of the base network using channel attention mechanisms and integrate the additive angular margin loss function (ArcFace loss) into the inner loop of MAML to constrain inner loop optimization and improve radar discrimination. Then, this network is used to classify small-sample micro-Doppler images obtained from millimeter-wave radar as the data source for pose recognition. Experimental tests were conducted on pose estimation and image classification tasks. The results demonstrate significant detection and recognition performance, with an accuracy of 94.5%, accompanied by a 95% confidence interval. Additionally, on the open-source dataset DIAT-µRadHAR, which is specially processed to increase classification difficulty, the network achieves a classification accuracy of 85.9%.

The Uncertainty Propagation for Carbon Atomic Interactions in Graphene under Resonant Vibration Based on Stochastic Finite Element Model.

Graphene is one of the most promising two-dimensional nanomaterials with broad applications in many fields. However, the variations and fluctuations in the material and geometrical properties are challenging issues that require more concern. In order to quantify uncertainty and analyze the impacts of uncertainty, a stochastic finite element model (SFEM) is proposed to propagate uncertainty for carbon atomic interactions under resonant vibration. Compared with the conventional truss or beam finite element models, both carbon atoms and carbon covalent bonds are considered by introducing plane elements. In addition, the determined values of the material and geometrical parameters are expanded into the related interval ranges with uniform probability density distributions. Based on the SFEM, the uncertainty propagation is performed by the Monte Carlo stochastic sampling process, and the resonant frequencies of graphene are provided by finite element computation. Furthermore, the correlation coefficients of characteristic parameters are computed based on the database of SFEM. The vibration modes of graphene with the extreme geometrical values are also provided and analyzed. According to the computed results, the minimum and maximum values of the first resonant frequency are 0.2131 and 16.894 THz, respectively, and the variance is 2.5899 THz. The proposed SFEM is an effective method to propagate uncertainty and analyze the impacts of uncertainty in the carbon atomic interactions of graphene. The work in this paper provides an important supplement to the atomic interaction modeling in nanomaterials.

A Kriging Surrogate Model for Uncertainty Analysis of Graphene Based on a Finite Element Method.

Due to the inevitable presence of random defects, unpredictable grain boundaries in macroscopic samples, stress concentration at clamping points, and unknown load distribution in the investigation of graphene sheets, uncertainties are crucial and challenging issues that require more exploration. The application of the Kriging surrogate model in vibration analysis of graphene sheets is proposed in this study. The Latin hypercube sampling method effectively propagates the uncertainties in geometrical and material properties of the finite element model. The accuracy and convergence of the Kriging surrogate model are confirmed by a comparison with the reported references. The uncertainty analysis for both Zigzag and Armchair graphene sheets are compared and discussed.