This paper presents a novel K-Band Quadrature Phase Shift Keying (QPSK) phase-modulated continuous-wave (PMCW) radar front end architecture with digital I/Q constellation phase compensation and I/Q channel imbalance c...
详细信息
Due to rapid urbanization and increasing energy use, Bangladesh is facing challenges in ensuring a reliable electricity supply, therefore it makes sensible to use more renewable resources in order to meet the rising e...
详细信息
This paper develops and tests a method for balancing capacitor voltages in 7-level flying-capacitor (FC) inverters, which are operated by the wavelet modulation (WM) technique. This multi-level inverter has two capaci...
详细信息
This paper develops and tests a method for balancing capacitor voltages in 7-level flying-capacitor (FC) inverters, which are operated by the wavelet modulation (WM) technique. This multi-level inverter has two capacitors in each pole, whose voltages deviate due to changes in its loading level. In order to ensure that all switching elements experience identical voltage stresses and no circulating currents, the voltage across each capacitor has to be maintained very close to its reference value. The proposed method to balance the capacitor voltage is based on adjusting the scales of resolution segmented wavelet basis functions, which are used as switching signals to operate a 7-level FC inverter. The adjustments of the scales can vary the widths and locations of switching pulses produced by the WM technique. The proposed capacitor voltage method is structured using a proportional-integral controller to adjust the scales, thus balancing the capacitor voltages. The accuracy, effectiveness, and response speed of the proposed method to balance capacitor voltage are demonstrated through simulation and experimental test cases. IEEE
Distributed detection (DD) plays a crucial role in sensor networks, where sensors gather data from a region of interest and report their observations to a fusion center (FC). The FC then makes a decision regarding a s...
详细信息
The undetected error probability of cyclic redundancy checks (CRC) combined with an error-correcting code in a concatenated coding scheme is considered. It is shown that this probability depends very much on the encod...
详细信息
Real-time traffic light detection and recognition (TLDR) remains a crucial challenge for autonomous vehicles (AVs), particularly in complex and chaotic traffic scenarios. These scenarios are defined by dense traffic, ...
详细信息
ISBN:
(纸本)9798331517786
Real-time traffic light detection and recognition (TLDR) remains a crucial challenge for autonomous vehicles (AVs), particularly in complex and chaotic traffic scenarios. These scenarios are defined by dense traffic, frequent occlusions, diverse vehicle types, and unpredictable movements, which can significantly impede traffic light detection. Current neural object detection models often struggle with such variability, leading to high false positive rates and reduced accuracy. This necessitates the development of novel models with advanced feature extraction, context awareness, and temporal reasoning to ensure safe AV operation. We present an end-to-end solution: 3D YOLO-SM (3D dept.-perception based on 2D object detection using YOLO, along with integrated State Machine), which features an enhanced object detector based on the YOLOv8 architecture and a Neural State Machine (NSM). Our approach incorporates dept. perception through stereo vision cameras to enhance detection accuracy, particularly for small and occluded traffic lights. We employ SPD-Convolution and attention mechanisms to improve feature learning capabilities, minimizing false positives by analyzing contextual information. The Neural Mealy Machine controller adjusts vehicle speed based on local traffic light detection, thereby imitating human driver behavior. Our model achieves state-of-the-art results on benchmark datasets, with mAP@0.5 scores of 92.5% on the LISA dataset and 89.3% on the Bosch dataset. In real-time tests conducted in urban traffic scenarios, the model demonstrated an accuracy of 97.38% with a processing time of 20 milliseconds per frame. Our proposed method surpasses existing state-of-the-art object detection models in detecting small and occluded traffic lights, especially in scenarios with weak semantic cues. This comprehensive approach ensures reliable and efficient traffic light recognition, significantly advancing autonomous vehicles' capabilities in navigating complex traffic e
Pneumonia, an inflammatory lung disease, impacts the small air sacs known as alveoli and is typically caused by the bacteria Streptococcus pneumonia. Symptoms commonly include a combination of productive or dry cough,...
详细信息
This paper presents a data-driven methodology that utilizes Dynamic Mode Decomposition (DMD) for the time-domain (TD) electromagnetic (EM) modeling of microwave devices. As an unsupervised machine learning technique, ...
详细信息
The nature of HF radio wave propagation through the lower ionosphere has significant impact on scientific efforts such as ELF/VLF wave generation and ionospheric sounding, as well as operational efforts e.g. over-the-...
详细信息
Braille script serves as a fundamental medium of communication for individuals with visual impairments. However, its interpretation by non-experts poses a significant challenge, hindering inclusivity in education and ...
详细信息
暂无评论