NASA's Planetary Data System (PDS) is a crucial repository for planetary mission data, yet its complex interface challenges users, especially non-experts. This research develops a framework that leverages OpenAI's GPT-4o-mini-API and a Retrieval-Augmented Generation (RAG) with the aim of simplifying user interaction with PDS. Using semantic embeddings, BM25 retrieval, and GPT-4o-mini, the system translates natural language queries into structured outputs, such as PDS compatible URLs. Our pilot implementation testing with simple queries achieved 95.24% accuracy in generating correct URLs, outperforming traditional TF-IDF with cosine similarity approach, which gained 90.91% accuracy. This work highlights the transformative potential of large language models to improve usability for scientific data repositories.
Limb loss and stroke-related motor impairments significantly impact individuals' mobility and independence, demanding advanced prosthetic and rehabilitation technologies. Surface electromyography has been widely used for prosthetic control, enabling gesture recognition through machine learning and deep learning models. However, sEMG-based systems often struggle with signal reliability and limited resolution for complex hand movements. Ultrasound has been explored as a complementary sensing modality, providing deeper muscle activity insights. This study proposes a novel deep learning approach integrating sEMG and US signals to enhance continuous 2-DoF hand motion estimation. We employ a sensor fusion technique utilizing an 8-channel sEMG armband and a handheld US probe, alongside an inertial measurement unit for ground truth data collection. Our method employs a hybrid convolutional neural network and long-short-term memory model to capture both spatial and temporal features of muscle activity, improving accuracy in predicting dynamic hand gestures. Experimental results suggest that the fusion of sEMG and US enhances continuous estimation accuracy particularly in fast and complex hand movements, compared to single-modality approach. The proposed method shows a potential of integrating hybrid sensing and deep learning for more intuitive and responsive prosthetic control systems.
We investigate a method to utilize source memory information for a Soft-Output Viterbi algorithm (SOVA) decoded Serially Concatenated Continuous Phase Modulation (SCCPM) system. The system concatenates a recursive systematic convolutional (RSC) code with a binary full-response continuous phase modulation (CPM) signal. We apply the SOVA as the Soft-Input Soft-Output (SISO) decoder for both the CPM and RSC components. We modify the RSC SOVA decoder to combine the a priori information provided by the CPM SOVA decoder with the a priori information from a Markov source. Simulation results show that incorporating source statistics information improves the error performance of the SCCPM system and accelerates decoding convergence during iterative decoding.
In today's digital age, a key challenge is the need for secure and reliable identity verification. Traditional methods, often relying on centralized authorities, are vulnerable to data breaches, privacy concerns, and inefficiency. For example, the 2017 Equifax breach exposed the personal data of over 147 million individuals, highlighting the risks of centralized data storage. Similarly, traditional identity verification methods in banking often involve lengthy processes prone to human error and fraud. These methods pose risks such as identity theft and lack of user control over personal data; for instance, studies show that identity theft affected over 42 million individuals in the U.S. in 2021 alone, highlighting the critical need for improved security measures. Therefore, we need to design a robust security and highly scalable framework to effectively address user privacy concerns. To address these challenges, we propose a blockchain-based digital identity verification system that leverages the immutability and transparency of blockchain technology. Our research provides a comprehensive discussion on the system's architecture, implementation, and potential challenges, paving the way for secure and decentralized identity verification frameworks. Our framework enhances data security, ensures user privacy, and fosters trust between stakeholders. The experiment's results demonstrate significant improvements in efficiency and enhanced security compared to centralized solutions.
Wheelchair users with limited or nonexistent lower-limb mobility often struggle with significant muscle atrophy, circulatory complications, and other health issues. Although exoskeletons for gait training have shown promise, many are optimized for upright ambulation and overlook individuals remaining fully seated due to clinical or practical constraints. This article presents a seat-focused exoskeleton synergy integrated with a wheelchair to facilitate multi-DOF lower-limb mobilization without requiring transfers or standing maneuvers. The system addresses a critical gap by providing repeated, lower-intensity knee and ankle exercises throughout the day-either passively, actively with assistance, or in a resistive training mode-thus promoting muscle engagement and circulation from a seated position. The exoskeleton leverages the wheelchair's power source and employs modular, adjustable rails to accommodate diverse leg lengths. Emphasis is placed on biomechanical alignment, safe torque control, and user comfort, supported by a model-based controller accounting for inertia, Coriolis effects, and gravity in real time. Bench tests and preliminary trials confirm stable torque output, accurate tracking of desired motion profiles, and minimal burden when attaching or detaching the device. The article introduces modeling for a two-DOF prototype (knee flexion/extension and ankle dorsiflexion/plantarflexion), expandable to include additional joints or degrees of freedom, enabling broader rehabilitation targets such as hip abduction/adduction. By embedding rehabilitative capabilities directly into the wheelchair environment, the proposed approach seeks to close the gap between infrequent clinic appointments and continuous therapy aimed at mitigating muscle ailments. A seat-focused exoskeleton designed for every-day use can facilitate accessible, on-demand rehabilitation and substantially improve long-term health outcomes for mobility impaired users.
Real-time object detection is pivotal in various computer vision and robotics applications, such as security surveillance, autonomous navigation, and interactive systems. This study evaluates several object detection algorithms under low illumination conditions, comparing traditional methods including SIFT, SURF, ORB, BRISK, KAZE, and AKAZE with the deep learning model YOLOv5. To simulate these conditions, we utilized the COCO dataset and implemented a sequence of preprocessing procedures, such as adjusting the gamma correction throughout the dataset, introducing Gaussian noise, and equalizing the histogram to improve image contrast. Here, YOLOv5 exhibited superior performance, achieving a detection accuracy of 92.47% and a processing time of 45 milliseconds per image. Traditional algorithms like SURF and KAZE, although slower, demonstrated precision rates of 78.54% and 76.86%, respectively. YOLOv5 also identified the highest number of key points and matches. Our findings indicate that while YOLOv5 excels in real-time detection under low-light conditions, traditional methods are effective for specific tasks in resource-constrained environments. This study benchmarks these algorithms and proposes a hybrid approach to enhance real-time object detection systems under low illumination conditions.
The fidelity and robustness of the models is critical for the efficacy of modern distribution system planning studies, serving as the foundation for strategic decision-making. However, systemic inconsistencies, incomplete datasets, and improper model data structuring introduce significant computational and operational complexities in power flow simulations, which are at the foundation of power system analyses and planning studies (such as hosting capacity evaluations). This paper presents a comprehensive analysis of prevalent inaccuracies and data anomalies that pervade distribution system models, categorizing them based on their impact on network topology, electrical parameter accuracy, and computational feasibility. The study investigates connectivity discrepancies, absent or misclassified asset characteristics, erroneous transformer-load associations, and data schema mismatches, elucidating their potential to distort system behavior and compromise reliability assessments. Furthermore, the paper highlights the imperative for adaptive data governance frameworks, advanced validation techniques, and automated rectification mechanisms to enhance distribution system model integrity.
Non-invasive pig weight prediction is crucial for efficient pork production and animal welfare monitoring. This study introduces a novel approach combining Segment Anything Model (SAM)-enhanced image preprocessing with deep learning techniques for accurate pig weight estimation using depth images. We compare traditional statistical methods with advanced deep learning models, focusing on minimizing preprocessing requirements and improving accuracy. Our custom image preprocessing pipeline, leveraging SAM for precise segmentation, enables efficient feature extraction and data cleaning. We evaluate various models, including Convolutional Neural Networks (CNNs), Multi-input CNNs, ResNets, and XceptionNet, against statistical methods such as linear regression, random forest, and support vector regression. Results demonstrate that deep learning models, particularly XceptionNet, significantly outperform statistical approaches. XceptionNet achieves a mean absolute error of 2.82 kg and a mean absolute percentage error of 7.42%, compared to the best statistical model (support vector regression) with 4.51 kg and 15.56%, respectively. This study highlights the potential of integrating state-of-the-art segmentation techniques like SAM with deep learning for enhanced accuracy in livestock monitoring, paving the way for real-time, non-invasive weight prediction systems in modern pig farming.
Wastewater treatment involves several processes that are difficult to accurately model and predict. Having accurate models of these processes is important for making control decisions, predicting treatment outcomes, and predicting the amount of energy used by the process. In recent literature, there have been several studies which investigate the possibility of using machine learning-based models for prediction or control of various wastewater treatment processes. One potential short-coming associated with many of these studies is that only the present or most recently measured value of influent water parameters were used to train these models. This study seeks to improve upon the use of machine learning models for predicting wastewater treatment processes by training machine learning models using time-series versions of the input variables. Instead of training the models to predict the future aeration intensity of a wastewater plant's air blowers based only upon present-time values of influent water parameters, the models are trained using a combination of the present values of the influent water parameters and a selected time range of historical values, called a sliding window. Using influent total suspended solids (TSS) and chemical oxygen demand (COD), five types of machine learning models are trained to predict the one-hour future aeration intensity of air blowers in a wastewater plant's aeration tank based on a window size ranging from the present value up to eight hours. The representative simulation-based case study results support that after an optimal window size is determined for each model, higher predictive accuracy can be attained than is possible using present values alone.
The segmentation of medical images, particularly for brain tumors, presents significant challenges due to the complexity, variability, and irregularity of tumor shapes and locations. Traditional methods struggle with accurately delineating tumor boundaries, often leading to inconsistent diagnoses and treatment outcomes. In this study, we propose an enhanced U-Net architecture designed to improve segmentation accuracy for brain tumor regions in magnetic resonance imaging (MRI) scans. The improved U-Net integrates advanced features such as residual connections and attention mechanisms, which help the model focus on critical tumor regions while addressing the challenges of vanishing gradients and complex patterns. Additionally, we employed techniques like multi-scale feature extraction to capture both fine-grained and global tumor details. The model was evaluated on a benchmark dataset, achieving significant improvements in metrics such as Dice coefficient and Intersection over Union (IoU), with scores consistently exceeding 0.92 and 0.90, respectively. Through rigorous training, evaluation, and the use of optimized loss functions, our proposed approach demonstrates robustness and accuracy in segmenting tumor regions, potentially paving the way for improved diagnostic precision and patient outcomes in clinical settings.
Rapid detection of dynamic blast events are critical for enhancing response strategies in both civilian and military applications. This paper introduces a groundbreaking ultra-low-latency blast detection system leveraging the unique capabilities of event cameras, which capture pixel-level changes in intensity at microsecond resolution. Unlike traditional video-based systems, this approach utilizes a real-time algorithm that processes events as they are captured, significantly reducing the detection latency. The proposed system combines histogram-based analysis with a novel diffusion model to accurately detect and characterize blast events through spatial and temporal event data aggregation. Experimental results demonstrate that this method achieves detection latencies on the order of milliseconds, substantially faster than conventional methods. The system's efficacy is validated through rigorous testing in simulated environments, showcasing its potential to significantly enhance real-time surveillance and safety monitoring systems. The implementation promises sub-stantial improvements in operational response times, crucial for environments where rapid reaction is paramount.
Music genre classification is a challenging problem due to the subjective and complex nature of music perception. This project explores three complementary approaches to genre classification: audio-based classification using Mel-spectrograms and Convolutional Neural Networks (CNN), lyrics-based classification using RoBERTa for natural language processing, and a multimodal classification approach that integrates both modalities. The audio-based approach achieved remarkable performance improvements through augmentation techniques, while the lyrics-based model demonstrated effective text comprehension using pre-trained transformers. Finally, a multimodal approach that combines Mel-spectrograms and lyrics through concatenation achieved the highest accuracy, underscoring the advantages of integrating diverse data modalities. These results demonstrate the potential of multimodal deep learning for complex classification tasks, opening the way to improved music retrieval systems.
Wafer defect detection is a critical aspect of semi-conductor manufacturing, where even minor defects can lead to significant performance issues or device failures. Traditional methods, such as rule-based approaches and Convolutional Neural Networks (CNNs), have achieved moderate success but face limitations in scalability, adaptability, and the ability to model complex spatial patterns in high-resolution wafer imagery. These limitations make it critical to investigate advanced architectures that can accurately capture complex spatial relationships. Vision Transformers (ViTs) have emerged as a promising alternative within computer vision, offering global attention techniques that are particularly effective in recognizing long-range dependencies. This study evaluates the feasibility of ViTs for wafer defect detection through a literature-based empirical validation approach, synthesizing findings from prior research on ViT-based classification, defect localization, and industrial feasibility. Results indicate that ViTs offer superior accuracy and interpretability compared to CNN s, particularly in detecting rare and mixed-type defects. However, challenges remain in terms of computational cost and real-time deployment in high-throughput manufacturing settings. The findings of this study underscore the potential of ViTs as a transformative technology for semiconductor defect inspection while highlighting key areas for further development.
The reliability and quality of 3D printing processes are critically dependent on the timely detection of mechanical faults. Traditional monitoring methods often rely on visual inspection and hardware sensors, which can be both costly and limited in scope. This paper explores a scalable and contactless method for the use of real-time audio signal analysis for detecting mechanical faults in 3D printers. By capturing and classifying audio data during the printing process, we aim to identify common faults such as nozzle clogging, filament breakage, pully skipping and various other mechanical faults. Utilizing Convolutional neural networks, we implement algorithms capable of real-time audio classification to detect these faults promptly. Our methodology involves conducting a series of controlled experiments to gather audio data, followed by the application of advanced machine learning models for fault detection. Additionally, we review existing literature on audio-based fault detection in manufacturing and 3D printing to contextualize our research within the broader field. Preliminary results demonstrate that audio signals, when analyzed with machine learning techniques, provide a reliable and cost-effective means of enhancing real-time fault detection.
Anti-reflection structured surfaces or coatings on large aperture optics need spatial transmission and reflection uniformity characterization. This work details the design and implementation of a linear mapping system built around a gantry for foot-scale optical windows.
While machine learning models perform well on offline data, assessing their performance in real-world, resource-constrained environments-considering accuracy, prediction time, power consumption, and memory usage-is crucial for practical applications. This research implements a mobile-based Human Activity Recognition solution to classify three postures-sitting, standing, and walking-using smartphone sensors, specifically accelerometer, gyroscope, and magnetometer. Time-domain features extracted from these sensors were used, with Random Forest employed for feature selection. One traditional machine learning model, Logistic Regression, and one deep learning model, Convolutional Neural Network, were trained and deployed via an Android application for real-time evaluation. While the Convolutional Neural Network achieved higher accuracy and better memory efficiency, Logistic Regression demonstrated faster prediction times during real-time use. Both models showed reduced accuracy for standing and walking postures in real-world conditions, emphasizing the challenges of deploying machine learning models in dynamic environments. This study highlights the importance of evaluating machine learning models in real-world settings to ensure reliability and efficiency, particularly in resource-constrained environments.
Traffic sign recognition is an important topic in the safety of connected and autonomous vehicles (CAVs). However, adversarial attacks pose significant threats by exploiting vulnerabilities in machine learning models to manipulate or misclassify traffic signs. This paper provides a comprehensive survey of adversarial attacks targeting traffic sign recognition systems. Specifically, it categorizes existing attack methodologies into distinct strategies, such as perturbation-based, patch-based, and targeted effects, offering a structured understanding of their approaches and implications. In addition, the paper reviews and organizes the current defense strategies used to mitigate these threats, highlighting their effectiveness and limitations. By presenting these insights, the survey aims to guide future research in developing robust and resilient traffic sign recognition systems to improve the safety of intelligent transportation systems.
Power line segmentation is a critical task in infrastructure monitoring, UAV-based inspection, and autonomous navigation. Detecting power lines in aerial images presents significant challenges due to their thin structure, varying backgrounds, and occlusions caused by environmental elements. Recent advancements in multimodal large language models (MLLMs) and vision-based segmentation frameworks provide a new direction for improving segmentation accuracy using natural language instructions. This study investigates the effectiveness of prompt engineering in enhancing power line segmentation performance within the Generalized Segmentation Vision Assistant (GSVA) framework. By systematically evaluating different prompt strategies, we analyze how varying levels of prompt specificity influence the balance between precision, recall, and segmentation completeness. Our findings demonstrate that carefully structured prompts improve the model's ability to detect power lines while minimizing false positives from visually similar structures. Furthermore, we propose an optimized prompt that enhances segmentation performance by leveraging contrast-awareness, structural positioning, and adaptive instruction tuning. The results highlight the importance of prompt optimization in improving segmentation accuracy, particularly for thin-object detection in complex backgrounds.
This study investigates the development of a Brain-Computer Interface (BCI) for controlling car functions using the MUSE headband, a consumer-grade EEG device. The primary objective is to map EEG signals to key driving commands, including acceleration, braking, and steering, with high accuracy and minimal latency. EEG signals were recorded from 5 participants performing cognitive tasks such as focus, relaxation, and deliberate eye blinks. Preprocessing involved band-pass filtering (1-50 Hz) and artifact removal through Independent Component Analysis (ICA). Features were extracted using Short-Time Fourier Transform (STFT) and classified using Fisher Linear Discriminant Analysis (LDA) and Convolutional Neural Networks (CNN). The results indicate that Fisher LDA achieved an average classification accuracy of 84.6%, while CNNs demonstrated superior performance with an accuracy of 91.3%. Real-time implementation exhibited an average command recognition latency of 320 ms. In simulation tests, the system successfully controlled acceleration, braking, and steering, achieving an overall execution accuracy of 87.8%. These findings demonstrate the feasibility of cost-effective EEG devices for real-time automotive BCI applications, offering significant potential for assistive driving technologies and hands-free vehicle control systems.
With the deployment of wide-area monitoring systems, power systems can be continuously monitored and analyzed, particularly during unusual events. Frequency fluctuations in the United States (US) interconnected power grid are influenced by various factors, including weather conditions and variability in solar power generation. This paper investigates the impact of the solar eclipse on April 8, 2024, on the frequency stability of the US power grid. Phasor measurement data from multiple interconnections are collected and analyzed. Methods of approximate frequency estimation for each interconnection are employed. The study involves comparing the frequency distribution, probability density, and variance on the eclipse day with those of non-eclipse days. The results provide insight into the grid's response to reduced solar generation and underscore the importance of predictive measures to maintain stability during solar eclipses and similar disruptions. The study shows that the Electric Reliability Council of Texas (ERCOT) experienced a dramatic increase in frequency fluctuations - 198% higher compared to periods without eclipse - highlighting the grid's vulnerability to reduced solar irradiance. Conversely, Eastern Interconnection (EI) demonstrated remarkable stability during the eclipse with minimal frequency fluctuations.
