Search Results (206)

Bird’s-eye-view (BEV) perception has emerged as a key representation for unified scene understanding in autonomous driving. However, current BEV methods relying solely on monocular cameras suffer from severe degradation under adverse weather and dynamic scenes due to limited depth cues and illumination dependency. To address these challenges, we propose a robust multi-modal BEV perception framework that integrates dual-source 4D millimeter-wave radar and multi-view camera images. The proposed architecture systematically exploits Doppler velocity and temporal information from 4D radar to model dynamic object motion, while introducing a deformable fusion strategy in the BEV space for accurate semantic alignment across modalities. Our design includes four key modules: a Doppler-Aware Radar Encoder (DARE) that enhances motion-sensitive features via velocity-guided attention; a Fog-Aware Feature Denoising Module (FADM) that suppresses modality inconsistency in low-visibility conditions through cross-modal attention and residual enhancement; a Multi-Modal Temporal Fusion Module (TFM) that encodes radar temporal sequences using a Transformer encoder for motion continuity modeling; and a confidence-aware multi-task loss that jointly supervises semantic segmentation, motion estimation, and object detection. Extensive experiments on the DualRadar dataset and adverse-weather simulations demonstrate that our method achieves significant gains over state-of-the-art baselines in BEV segmentation accuracy, detection robustness, and motion stability. The proposed framework offers a scalable and resilient solution for real-world autonomous perception, especially under challenging environmental conditions. Full article

Multiple-input multiple-output (MIMO) radar is widely adopted in the fields of forward-looking imaging and target recognition, but its azimuth imaging resolution is fundamentally limited by the size of the physical aperture. Aiming to achieve higher imaging resolution than the theoretical value, an image super-resolution reconstruction method based on the horizontal attention generative adversarial network (HA-GAN) is proposed in this paper. In detail, the horizontal attention mechanism is introduced into the generator to enhance the azimuth resolution, and then the high-resolution (HR) images can be obtained through the adversarial learning between the generator network and the discriminator network. The numerical results demonstrate that the proposed method can break through the theoretical limitation of MIMO azimuth imaging. Moreover, compared to some state-of-the-art methods, the proposed method demonstrates superior performance on sidelobe suppression and super-resolution reconstruction at a low signal-to-noise ratio (SNR). Furthermore, the method’s effectiveness and generalization capability are extensively validated using simulation data, real-world experiments on a millimeter-wave MIMO system, and the public CRUW and RADAL datasets. Overall, the experimental results demonstrate that HA-GAN significantly enhances angular resolution and target recoverability, establishing it as a robust solution for high-precision forward-looking radar imaging. Full article

Terahertz (THz) sub-millimeter wave imaging offers unique capabilities for stand-off biometrics through concealment, yet it suffers from severe sparsity, low resolution, and high noise. To address these limitations, we introduce a novel unified Multi-Task Learning (MTL) network centered on a custom shared U-Net-like THz data encoder. This network is designed to simultaneously solve three distinct critical tasks on concealed THz facial data, given a limited dataset of approximately 1400 THz facial images of 20 different identities. The tasks include concealed face verification, facial posture classification, and a generative reconstruction of unconcealed faces from concealed ones. While providing highly successful MTL results as a standalone solution on the very challenging dataset, we further studied the expansion of this architecture via a cross-modal teacher-student approach. During training, a privileged visible-spectrum teacher fuses limited visible features with THz data to guide the THz-only student. This distillation process yields a student network that relies solely on THz inputs at inference. The cross-modal trained student achieves better latent space in terms of inter-class separability compared to the single-modality baseline, but with reduced intra-class compactness, while maintaining a similar success in the task performances. Both THz-only and distilled models preserve high unconcealed face generative fidelity. Full article

Terahertz waves are located in the “transition zone” between millimeter waves and infrared light. Terahertz video synthetic aperture radar (THz-ViSAR) utilizes the high operating frequency, strong radar cross-section intensity, and high azimuth repetition frequency of terahertz waves to detect and track ground moving targets. The conventional methods for detecting moving targets do not take into account the imaging characteristics of moving targets in THz-ViSAR. The constant false alarm rate (CFAR) detection method is used together with other methods to detect moving targets, resulting in unsatisfactory detection performance. This article proposes a new detection method for single channel slow-moving targets in THz-ViSAR based on shadows and light spots, which extracts the features of the shadow and spot areas of the moving target, and determines the position and direction of the moving target through the identification of the shadow and spot areas. The progressiveness of this method is verified by simulation and experimental tests. Full article

To address the critical demand for high-gain, low-sidelobe, and high-efficiency antennas in W-band arrays, this work presents a low-sidelobe all-metal array antenna based on ridge gap waveguide technology. The design employs a three-layer contactless metal structure, integrating a stepped-ridge feeding network with Taylor amplitude distribution and a higher-order mode resonant cavity. This integration enables efficient power distribution and low-loss transmission while eliminating the need for conventional welding or bonding processes. Measurement results indicate that the antenna exhibits a reflection coefficient below −10 dB across the 92.5–103.5 GHz. The in-band gain exceeds 25.8 dBi with less than 1 dB fluctuation, and the radiation efficiency surpasses 78%. Specifically, the sidelobe levels in both E- and H-planes remain below −17.5 dB, reaching under −19.5 dB at 94 GHz, while cross-polarization is better than −30 dB. The proposed antenna demonstrates high gain, low sidelobe, and high efficiency, showing promising potential for applications in millimeter-wave radar, imaging, and 6G communication systems. Full article

Fifth-generation (5G) wireless networks have been widely deployed across various applications, including indoor positioning. This paper presents a model for 3D indoor localization of an unmanned aerial vehicle (UAV) using 5G millimeter-wave technology. Wireless InSite software is used to simulate a real-world environment and extract channel state information from multiple 5G next-generation NodeBs (gNBs), which is then used to generate channel frequency response (CFR) images. These images are employed in a fingerprinting method, where a deep convolutional neural network is trained for accurate position prediction. The model is trained across multiple scenarios involving changes in the number of gNBs, receiver positions, and spacing. In all scenarios, the model is tested using a UAV flying along a trajectory at variable speed. It is shown that a mean positioning error (MPE) of 0.36 m in 2D and 0.43 m in 3D is achieved when twelve gNBs with receivers spaced at 0.25 m are used. In addition, the corresponding root mean square error (RMSE) values of 0.32 m (2D) and 0.33 m (3D) further confirm the stability of the localization performance by indicating a low dispersion of positioning errors. This demonstrates that high positioning accuracy is feasible, even when synchronization errors and hardware imperfections exist. Full article

Hand gesture recognition (HGR) systems play a critical role in modern intelligent transportation frameworks by enabling reliable communication between pedestrians, traffic operators, and autonomous vehicles. This work presents a novel traffic hand gesture recognition method that combines nine grayscale radar images captured from multiple millimeter-wave radar nodes into a single RGB representation through an optimized rotation–shift fusion strategy. This transformation preserves complementary spatial information while minimizing inter-image interference, enabling deep learning models to more effectively utilize the distinctive micro-Doppler and spatial patterns embedded in radar measurements. Extensive experimental studies were conducted to verify the model’s performance, demonstrating that the proposed RGB fusion approach provides higher classification accuracy than single-sensor or unfused representations. In addition, the proposed model outperformed state-of-the-art methods in the literature with an accuracy of 92.55%. These results highlight its potential as a lightweight yet powerful solution for reliable gesture interpretation in future intelligent transportation and human–vehicle interaction systems. Full article

Microwave, millimeter-wave, and terahertz devices are fundamental to modern 5G/6G communications, automotive imaging radar, and sensing systems. As essential RF front-end elements, passive microwave array components on static platforms remain constrained by fixed geometry and single-frequency excitation, leading to limited spatial resolution and weak interference suppression. Phase-steered arrays offer angular control but lack range-dependent response, preventing true two-dimensional focusing. Frequency-Diverse Array Multiple-Input Multiple-Output (FDA-MIMO) architectures introduce element-wise frequency offsets to enrich spatial–spectral degrees of freedom, yet conventional linear or predetermined nonlinear offsets cause range–angle coupling, periodic lobes, and restricted beamforming flexibility. Existing optimization strategies also tend to target single objectives and insufficiently address target- or scene-induced perturbations. This work proposes a nonlinear frequency-offset design for passive microwave arrays using a Dingo–Gray Wolf hybrid intelligent optimizer. A multi-metric fitness function simultaneously enforces sidelobe suppression, null shaping, and frequency-offset smoothness. Simulations in static scenarios show that the method achieves high-resolution two-dimensional focusing, enhanced interference suppression, and stable performance under realistic spatial–spectral mismatches. The results demonstrate an effective approach for improving the controllability and robustness of passive microwave array components on static platforms. Full article

Distributed fiber-optic photoacoustic non-destructive testing (DFP-NDT) represents a paradigm shift from passive sensing to active probing, fundamentally transforming structural health monitoring through integrated fiber-based ultrasonic generation and detection capabilities. This review systematically examines DFP-NDT’s evolution by following the technology’s natural progression from fundamental principles to practical implementations. Unlike conventional approaches that require external excitation mechanisms, DFP-NDT leverages photoacoustic transducers as integrated active components where fiber-optical devices themselves generate and detect ultrasonic waves. Central to this technology are photoacoustic materials engineered to maximize conversion efficiency—from carbon nanotube-polymer composites achieving 2.74 × 10⁻² conversion efficiency to innovative MXene-based systems that combine high photothermal conversion with structural protection functionality. These materials operate within sophisticated microstructural frameworks—including tilted fiber Bragg gratings, collapsed photonic crystal fibers, and functionalized polymer coatings—that enable precise control over optical-to-thermal-to-acoustic energy conversion. Six primary distributed fiber-optic photoacoustic transducer array (DFOPTA) methodologies have been developed to transform single-point transducers into multiplexed systems, with low-frequency variants significantly extending penetration capability while maintaining high spatial resolution. Recent advances in imaging algorithms have particular emphasis on techniques specifically adapted for distributed photoacoustic data, including innovative computational frameworks that overcome traditional algorithmic limitations through sophisticated statistical modeling. Documented applications demonstrate DFP-NDT’s exceptional versatility across structural monitoring scenarios, achieving impressive performance metrics including 90 × 54 cm² coverage areas, sub-millimeter resolution, and robust operation under complex multimodal interference conditions. Despite these advances, key challenges remain in scaling multiplexing density, expanding operational robustness for extreme environments, and developing algorithms specifically optimized for simultaneous multi-source excitation. This review establishes a clear roadmap for future development where enhanced multiplexed architectures, domain-specific material innovations, and purpose-built computational frameworks will transition DFP-NDT from promising laboratory demonstrations to deployable industrial solutions for comprehensive structural integrity assessment. Full article

Aiming at the problems of insufficient environmental perception capability of autonomous mobile robots and low multi-modal data fusion efficiency in the complex underground coal mine environment featuring low illumination, high dust, and dynamic obstacles, a reliable passable region identification method for autonomous mobile robots operating in underground coal mine is proposed in this paper. Through the spatial synchronous installation strategy of dual 4D millimeter-wave radars and dynamic coordinate system registration technology, it increases point cloud density and effectively enhances the spatial characterization of roadway structures and obstacles. Combining the characteristics of infrared thermal imaging and the penetration advantage of millimeter-wave radar, a multi-modal data complementary mechanism based on decision-level fusion is proposed to solve the perceptual blind zones of single sensors in extreme environments. Integrated with lightweight model optimization and system integration technology, an intelligent environmental perception system adaptable to harsh working conditions is constructed. The experiments were carried out in the simulated tunnel. The experiments were carried out in the simulated tunnel. The experimental results indicate that the robot can utilize the data collected by the infrared camera and the radar to identify the specific distance to obstacles, and can smoothly achieve the recognition and marking of passable areas. Full article

Time-division duplex (TDD) transceivers have found broad utility in millimeter-wave 5G communication, radar and imaging applications. The co-design of the switch and transmit/receive (T/R) amplifiers becomes essential in optimizing the passive loss and chip size. This work presents a Ka-band T/R amplifier with an embedded switch topology. The amplification cores from the TX and RX channels reuse the matching network to the T/R common port, and the full combination of switching and matching structures is enabled within a compact two-winding transformer. Implemented in 40 nm CMOS technology, the proof-of-concept Ka-band T/R amplifier occupies a core chip area of 0.163 mm². Experimental results show that it achieves a peak gain of 17.2 dB with a −3 dB bandwidth of 22.6–30.2 GHz in TX mode and a peak of 17.1 dB with a −3 dB bandwidth of 23.4–31.0 GHz in RX mode. The compact size and wideband gain response make the proposed T/R amplifier suitable for Ka-band TDD applications. Full article

This letter presents a high-gain frequency-controlled beam-scanning antenna specifically designed for through-wall radar (TWR) applications in the W band. The antenna leverages the leaky-wave radiation generated by spoof surface plasmon polaritons (SSPPs) propagating on sinusoidally modulated reactance surfaces (SMRS). Periodically arranged quasi-H-shaped metallic cells are employed to achieve beam scanning. The integration of a flared structure at the apex of the designed SSPP antenna results in a significant gain enhancement, yielding an approximate increase of 10 dB. From 92.8 to 97.6 GHz, the antenna exhibits a reflection coefficient of |S11| < −10 dB, provides a high scanning rate of 4.05°/%, and achieves a realized gain of 20.9 dBi. This design eliminates the necessity for mechanical rotators and phase shifters that are typical in traditional TWR systems, significantly reducing system complexity and cost. A vehicle-mounted W-band TWR system was developed, integrating the designed SSPP antenna and employing linear frequency modulation technology to emit millimeter-wave signals for electronic scanning detection. With an economical and efficient design approach, testing has demonstrated that the system can perform through-wall imaging at a distance of 10 m, both in stationary and in motion conditions. Full article

This paper presents a compact 122 GHz Terahertz FMCW radar using a plastic hollow-core dielectric waveguide for non-destructive testing. The guided approach simplifies the system, avoiding complex free-space optics and alignment, while improving the signal-to-noise ratio by isolating endpoint reflections from internal ones. Various configurations, including solid immersion lenses, enhance spatial resolution and imaging capabilities. Experiments combine 3D electromagnetic simulations and raster scanning to image fine details and detect subsurface defects. Applications span aerospace, automotive, and art conservation. Results demonstrate that the guided FMCW radar is a cost-effective, portable, and reliable alternative to traditional free-space setups, enabling broader, practical implementation across industries. Full article

While recent industrial automation trends emphasize the importance of non-destructive inspection by material-identifying millimeter-wave, terahertz-wave, and infrared (MMW, THz, IR) monitoring, fundamental tools in these wavelength bands (such as sensors) are still immature. Although inorganic semiconductors serve as diverse sensors with well-established large-scale fine-processing fabrication, the use of those devices is insufficient for non-destructive monitoring due to the lack of photo-absorbent properties for such major materials in partial regions across MMW–IR wavelengths. To satisfy the inherent advantageous non-destructive MMW–IR material identification, ultrabroadband operation is indispensable for photo-sensors under compact structure, flexible designability, and sensitive performances. This review then introduces the recent advances of carbon nanotube film-based photo-thermoelectric imagers regarding usable and high-yield device fabrication techniques and scientific synergy among computer vision to collectively satisfy material identification with three-dimensional (3D) structure reconstruction. This review synergizes material science, printable electronics, high-yield fabrication, sensor devices, optical measurements, and imaging into guidelines as functional non-destructive inspection platforms. The motivation of this review is to introduce the recent scientific fusion of MMW–IR sensors with visible-light computer vision, and emphasize its significance (non-invasive material-identifying sub-millimeter-resolution 3D-reconstruction with 660 nm–1.15 mm-wavelength imagers at noise equivalent power within 100 pWHz^−1/2) among the existing testing methods. Full article

This study presents MDSCNet, a compact radar image-based deep learning model for multi-action classification in elderly healthcare scenarios. Motivated by the need for real-time deployment on resource-constrained devices, MDSCNet employs a streamlined architecture with a small number of lightweight expansion–depthwise–projection blocks, removing complex attention and squeeze-and-excitation modules to minimize computational overhead. The model is evaluated on a millimeter-wave radar dataset covering five healthcare-related actions: lying, sitting, standing, bed-exit, and falling, performed by 15 participants on an actual electric nursing bed. The experimental results demonstrate that MDSCNet achieves accuracy comparable to state-of-the-art CNN-based methods while maintaining an extremely compact model size of only 0.29 MB, showing its suitability for practical elderly care applications where both accuracy and efficiency are critical. Full article