Search

The development of safety-critical real-time embedded systems (RTESs) is a difficult task. Indeed, in addition to functional requirements, these systems must guarantee timing constraints. Model-driven engineering (MDE) promotes a rise in the level of abstraction during the development process of these systems. This is beneficial since it may help designers manage the increasing complexity of these systems. It may also accelerate the development process through model transformations and enable early verification of the timing properties. Schedulability analysis techniques aim to verify whether the tasks complete their executions within the time limit specified by the real-time application (i.e., the deadline). Unfortunately, software designers find this task difficult since it requires a deep understanding of real-time scheduling theory, as well as the use of dedicated tools (called analysis tools). In this paper, we propose real-time verification design patterns to promote early verification of timing properties in a model-based approach. These patterns aim to assist designers in the verification process. They encapsulate the knowledge of a real-time expert in scheduling theory. To facilitate and fasten the verification process, we propose to automate the generation of the analysis model. Thanks to the provided patterns, this generation phase includes an automatic consistency check of the design model. The consistency check step verifies the design's validity and completeness with respect to the applied pattern. To add further benefits to the proposed patterns, we describe a process that provides context for how these patterns can be applied. We apply and simulate the contribution to real-world case study. The obtained results show the viability of this research. [ABSTRACT FROM AUTHOR]

Unmanned Aerial Vehicle (UAV) infrared detection has problems such as weak and small targets, complex backgrounds, and poor real-time detection performance. It is difficult for general target detection algorithms to achieve the requirements of a high detection rate, low missed detection rate, and high real-time performance. In order to solve these problems, this paper proposes an improved small target detection method based on Picodet. First, to address the problem of poor real-time performance, an improved lightweight LCNet network was introduced as the backbone network for feature extraction. Secondly, in order to solve the problems of high false detection rate and missed detection rate due to weak targets, the Squeeze-and-Excitation module was added and the feature pyramid structure was improved. Experimental results obtained on the HIT-UAV public dataset show that the improved detection model's real-time frame rate increased by 31 fps and the average accuracy (MAP) increased by 7%, which proves the effectiveness of this method for UAV infrared small target detection. [ABSTRACT FROM AUTHOR]

Energy management is important for both consumers and utility providers. Utility providers are concerned with identifying and reducing energy wastage and thefts. Consumers are interested in reducing their energy consumption and bills. In Pakistan, residential and industrial estates account for nearly 31,000 MW of the maximum total demand, while the transmission and distribution capacity has stalled at about 22,000 MW. This 9000 MW gap in demand and supply, as reported in 2022, has led to frequent load shedding. Although the country now has an excess generation capacity of about 45,000 MW, the aging transmission and distribution network cannot deliver the requisite power at all times. Hence, electricity-related problems are likely to continue for the next few years in the country and the same is true for other low- and middle-income countries (LMICs). Several energy monitoring systems (EnMS) have been proposed, but they face limitations in terms of cost, ease of application, lack of universal installation capability, customization, and data security. The research below focused on the development of an economical, secure, and customizable real-time EnMS. The proposed EnMS comprises low-cost hardware for gathering energy data with universal compatibility, a secured communication module for real-time data transmission, and a dashboard application for visualization of real-time energy consumption in a user-preferred manner, making the information easily accessible and actionable. The experimental results and analysis revealed that approximately 40% cost savings in EnMS development could be achieved compared to other commercially available EnMSs. The performance of the EnMS hardware was evaluated and validated through rigorous on-site experiments. The front-end of the EnMS was assessed through surveys and was found to be interactive and user-friendly for the target clients. The developed EnMS architecture was found to be an economical end-product and an appropriate approach for small and medium clients such as residential, institutional, commercial, and industrial consumers, all on one platform. [ABSTRACT FROM AUTHOR]

Summary: Due to recent developments in location‐based services and mobile computing, the need for indices for moving objects has been strengthened to improve the response time of a query operation. With a single index in place for managing both the update and query operations for moving objects, the index needs to be updated each time the object moves. This deteriorates the performance of concurrent query operations. It is critical to handle the conflicts between continuous update and query operations effectively using appropriate concurrency control protocol otherwise, inconsistent results will be reported. Many indices have been proposed in the literature for moving objects but they lack the support for processing concurrent operations. Further, the consistent indices in the literature are based on tree structure that have computationally expensive split/merge operations that can negatively affect the response time of query processing algorithms. Moreover such tree based indices are proposed for historical and future timeline data. As the scope of this article is on current timeline and concurrent continuous query operations, therefore, we exploit state of the art hash‐based indices in the literature and presented the two consistent versions. The comparative analysis of the indices is performed and meaningful findings are also presented along. [ABSTRACT FROM AUTHOR]

In this brief, a real-time inversion of petrophysical information in downhole firmware for nuclear magnetic resonance logging while drilling (NMR-LWD) based on a field-programmable gate array (FPGA) is proposed. Firstly, the inverse problem is introduced, the nonnegative least-squares (NNLS) problem is modeled and solved by the accelerated projected gradient descent (APGD) algorithm. Secondly, the efficient architecture for the APGD accelerator based on look-up tables (LUTs) is used in Intel's 10M50 FPGA. Finally, a software cooling method based on a multi-core structure and periodic thermal management technology is deployed. The proposed scheme is applied to NMR-LWD, the effectiveness and feasibility of the proposed scheme are verified experimentally. [ABSTRACT FROM AUTHOR]

In limit conditions, autonomous vehicles face the risk of lateral instability. The integrated control of steering and braking, an important measure for improving the stability of autonomous vehicles, has been extensively studied. A novel steering and braking integrated model predictive path tracking control (PTC) based on a dimension reduction model is proposed in this study. This method aims at the dilemma of the real-time limitation of the current integrated model predictive PTC based on nonlinear vehicle dynamics in practical applications and the unsatisfactory control effect of the integrated model predictive PTC based on the linearised vehicle dynamics in limit conditions. The core concept of this study is to reduce the input dimension of the integrated controller model by designing a model dimension reduction method, thereby reducing the decision variables of the optimisation problem and improving the real-time performance. The model dimension reduction method is designed based on the optimal utilisation of tire force to ensure the control performance of the proposed integrated control method in limit conditions. The integrated control method based on the dimension reduction model is compared with several existing integrated control methods in limit conditions to demonstrate its validity and superiority. The simulation tests with Simulink and CarSim indicate that the proposed method can reduce the calculation time by more than 40 $ \% $ % on the premise of ensuring path tracking accuracy and vehicle stability in limit conditions. Moreover, the hardware-in-the-loop tests prove the practicability of the presented method. [ABSTRACT FROM AUTHOR]

Currently, numerous high-precision models have been proposed for semantic segmentation, but the model parameters are large and the segmentation speed is slow. Real-time semantic segmentation for urban scenes necessitates a balance between accuracy, inference speed, and model size. In this paper, we present an efficient solution to this challenge, efficient asymmetric attention module net (EAAMNet) for the semantic segmentation of urban scenes, which adopts an asymmetric encoder–decoder structure. The encoder part of the network utilizes an efficient asymmetric attention module to form the network backbone. In the decoding part, we propose a lightweight multi-feature fusion decoder that can maintain good segmentation accuracy with a small number of parameters. Our extensive evaluations demonstrate that EAAMNet achieves a favorable equilibrium between segmentation efficiency, model parameters, and segmentation accuracy, rendering it highly suitable for real-time semantic segmentation in urban scenes. Remarkably, EAAMNet attains a 73.31% mIoU at 128 fps on Cityscapes and a 69.32% mIoU at 141 fps on CamVid without any pre-training. Compared to state-of-the-art models, our approach not only matches their model parameters but also enhances accuracy and increases speed. [ABSTRACT FROM AUTHOR]

Analysis of cutting phenomena has been conducted for a long time and many researchers have elucidated the phenomena that may occur at the cutting-edge during machining; for example, built-up edges and welding phenomena on cutting edge. However, the existing research has focused on observing the tool and chips after machining, when the tool and work material have been cooled, and are under atmospheric pressure. Therefore, we consider that it is different from the phenomenon that occurs during cutting. Because the cutting edge of a tool is in a high-temperature and high-pressure environment during machining and is released from such an environment after machining, the cutting edge must be visualized to discuss the actual cutting phenomena. There have been research reports in the past that have visualized the backside of cutting chips; however, that experiment was far from actual cutting phenomena and was conducted at significantly low cutting speed. Thus, the visualization of the cutting phenomenon is physically extremely difficult at the cutting edge. Under such circumstances, this study developed a device that observed the behavior of the boundary surface between cutting chips and the rake face in real time from the backside of the rake face during machining. By using the developed device, we could visualize this phenomenon to acquire the data directly and visually, which can otherwise be grasped only indirectly. In this device, a camera was mounted on a tool holder, and a cutting tip made of a transparent material was used to observe rake face, cutting edge, and flank face which can be visualized during machining from the backside of the rake face within one field of view. This paper reports on the outline of the developed device, its method, and the results of experimental observation of the state of two-dimensional cutting using a lathe. [ABSTRACT FROM AUTHOR]

Reconstructing the three‐dimensional (3D) shape and texture of the face from a single image is a significant and challenging task in computer vision and graphics. In recent years, learning‐based reconstruction methods have exhibited outstanding performance, but their effectiveness is severely constrained by the scarcity of available training data with 3D annotations. To address this issue, we present the PR3D (Precise and Realistic 3D face reconstruction) method, which consists of high‐precision shape reconstruction based on semi‐supervised learning and high‐fidelity texture reconstruction based on StyleGAN2. In shape reconstruction, we use in‐the‐wild face images and 3D annotated datasets to train the auxiliary encoder and the identity encoder, encoding the input image into parameters of FLAME (a parametric 3D face model). Simultaneously, a novel semi‐supervised hybrid landmark loss is designed to more effectively learn from in‐the‐wild face images and 3D annotated datasets. Furthermore, to meet the real‐time requirements in practical applications, a lightweight shape reconstruction model called fast‐PR3D is distilled through teacher–student learning. In texture reconstruction, we propose a texture extraction method based on face reenactment in StyleGAN2 style space, extracting texture from the source and reenacted face images to constitute a facial texture map. Extensive experiments have demonstrated the state‐of‐the‐art performance of our method. [ABSTRACT FROM AUTHOR]

This paper presents an implementation of an architecture based on opensource solutions using ELK Stack – Elasticsearch, Logstash, and Kibana – for realtime data analysis and visualizations in the Medical Data Integration Center, University Hospital Cologne, Germany. The architecture addresses challenges in handling diverse data sources, ensuring standardized access, and facilitating seamless analysis in real-time, ultimately enhancing the precision, speed, and quality of monitoring processes within the medical informatics domain. [ABSTRACT FROM AUTHOR]

Purpose: To develop a highly‐accelerated, real‐time phase contrast (rtPC) MRI pulse sequence with 40 fps frame rate (25 ms effective temporal resolution). Methods: Highly‐accelerated golden‐angle radial sparse parallel (GRASP) with over regularization may result in temporal blurring, which in turn causes underestimation of peak velocity. Thus, we amplified GRASP performance by synergistically combining view‐sharing (VS) and k‐space weighted image contrast (KWIC) filtering. In 17 pediatric patients with congenital heart disease (CHD), the conventional GRASP and the proposed GRASP amplified by VS and KWIC (or GRASP + VS + KWIC) reconstruction for rtPC MRI were compared with respect to clinical standard PC MRI in measuring hemodynamic parameters (peak velocity, forward volume, backward volume, regurgitant fraction) at four locations (aortic valve, pulmonary valve, left and right pulmonary arteries). Results: The proposed reconstruction method (GRASP + VS + KWIC) achieved better effective spatial resolution (i.e., image sharpness) compared with conventional GRASP, ultimately reducing the underestimation of peak velocity from 17.4% to 6.4%. The hemodynamic metrics (peak velocity, volumes) were not significantly (p > 0.99) different between GRASP + VS + KWIC and clinical PC, whereas peak velocity was significantly (p < 0.007) lower for conventional GRASP. RtPC with GRASP + VS + KWIC also showed the ability to assess beat‐to‐beat variation and detect the highest peak among peaks. Conclusion: The synergistic combination of GRASP, VS, and KWIC achieves 25 ms effective temporal resolution (40 fps frame rate), while minimizing the underestimation of peak velocity compared with conventional GRASP. [ABSTRACT FROM AUTHOR]

Software-Defined Networking (SDN) is a modern networking architecture that segregates control logic from data plane and supports a loosely coupled architecture. It provides flexibility in this advanced networking paradigm for any changes. Further, it controls the complete network in a centralized using controller(s). However, it comes with several security issues: Exhausting bandwidth and flow tables, Distributed Denial of Service (DDoS) attacks, etc. DDoS is a powerful attack for Internet-based applications and services, traditional and SDN paradigms. In the case of the SDN environment, attackers frequently target the central controller(s). This paper proposes a Kafka Streams-based real-time DDoS attacks classification approach for the SDN environment, named KS-SDN-DDoS. The KS-SDN-DDoS has been designed using highly scalable H2O ML techniques on the two-node Apache Hadoop Cluster (AHC). It consists of two modules: (i) Network Traffic Capture (NTCapture) and (ii) Attack Detection and Traffic Classification (ADTClassification). The NTCapture is deployed on the two nodes Apache Kafka Streams Cluster (AKSC-1). It captures incoming network traffic, extracts and formulates attributes, and publishes significant network traffic attributes on the Kafka topic. The ADTClassification is deployed on the two nodes Apache Kafka Streams Cluster (AKSC-2). It consumes network flows from the Kafka topic, classifies it based on the ten attributes, and publishes it to the decision Kafka topic. Further, it saves attributes with outcome to the Hadoop Distributed File System (HDFS). The KS-SDN-DDoS approach is designed and validated using the recent “DDoS Attack SDN dataset”. The result shows that the proposed system gives better classification accuracy (100%). [ABSTRACT FROM AUTHOR]

In this paper, the implementation of a new pupil detection system based on artificial intelligence techniques suitable for real-time and real-word applications is presented. The proposed AI-based pupil detection system uses a classifier implemented with slim-type neural networks, with its classes being defined according to the possible positions of the pupil within the eye image. In order to reduce the complexity of the neural network, a new parallel architecture is used in which two independent classifiers deliver the pupil center coordinates. The training, testing, and validation of the proposed system were performed using almost 40,000 eye images with a resolution of 320 × 240 pixels and coming from 20 different databases, with a high degree of generality. The experimental results show a detection rate of 96.29% at five pixels with a standard deviation of 3.38 pixels for all eye images from all databases and a processing speed of 100 frames/s. These results indicate both high accuracy and high processing speed, and they allow us to use the proposed solution for different real-time applications in variable and non-uniform lighting conditions, in fields such as assistive technology to communicate with neuromotor-disabled patients by using eye typing, in computer gaming, and in the automotive industry for increasing traffic safety by monitoring the driver's cognitive state. [ABSTRACT FROM AUTHOR]

The utilization of Low Earth Orbit (LEO) satellites is anticipated to augment various aspects of traditional GNSS-based Positioning, Navigation, and Timing (PNT) services. While the LEO satellite orbital products can nowadays be produced with rather high accuracy in real-time of a few centimeters, the precision of the LEO satellite clock products that can be achieved in real-time is less studied. The latter, however, plays an essential role in the LEO-augmented positioning and timing performances. In real-time, the users eventually use the predicted LEO satellite clocks, with their precision determined by both the near-real-time clock precision and the prediction time needed to match the time window for real-time applications, i.e., the precision loss during the prediction phase. In this study, a real-time LEO satellite clock determination method, consisting of near-real-time clock determination with ultra-short-term clock prediction is proposed and implemented. The principles and strategies of this method are discussed in detail. The proposed method utilized Kalman-filter-based processing, but supports restarts at pre-defined times, thus hampering continuous bias propagation and accumulation from ancient epochs. Based on the method, using Sentinel-3B GNSS observations and the real-time GNSS products from the National Center for Space Studies (CNES) in France, the near-real-time LEO satellite clocks can reach a precision of 0.2 to 0.3 ns, and the precision loss during the prediction phase is within 0.07 ns for a prediction time window from 30 to 90 s. This results in a total error budget in the real-time LEO satellite clocks of about 0.3 ns. [ABSTRACT FROM AUTHOR]

The delivery of concrete is a crucial process in construction projects, and any delay or error can cause significant setbacks and added costs. Thus, effective real-time management of concrete delivery is essential to ensure timely and successful project completion. In this paper, we will discuss a practical and manufacturer-agnostic approach to real-time management of concrete delivery for construction named BET 4.0 that is being conceived with a close partnership with a construction company. This application provides the possibility to optimize the whole concreting process as it establishes the connection between all the relevant components and stakeholders involved in the construction process, namely the concrete plant, the transport, and the construction site, interfacing with all actors involved, and benefiting from real-time data produced by installed sensors in the several components such as machines, plants, or construction elements. [ABSTRACT FROM AUTHOR]

Diastolic/systolic blood pressure ratio (D/S) ≥ 1.2 is the gold standard of enhanced external counterpulsation (EECP) treatment, but it does not show a clear clinical correspondence with the configuration of the EECP mode. As such, a single target results in different treatment effects in different individuals. The local haemodynamic effect (wall shear stress, WSS) of EECP on vascular endothelial cells is conducive to promote the growth of collateral circulation vessels and restore the blood supply distal to the stenosis lesion. Considering the haemodynamic effects of WSS on human arteries, this study developed a real‐time patient‐specific treatment strategy of EECP for patients with cardio‐cerebrovascular diseases. Based on patient‐specific haemodynamic data from 113 individuals, an optimization algorithm was developed to achieve the individualization of a 0D lumped‐parameter model of the human circulatory system, thereby simulating the patient‐specific global haemodynamic effects. 0D/3D coupled cardio‐cerebrovascular models of two subjects were established to simulate the local WSS. We then established statistical models to evaluate clinically unmeasurable WSS based on measurable global haemodynamic indicators. With the aim of attaining appropriate area‐ and time‐averaged WSS (ATAWSS, 4–7 Pa), as evaluated by global haemodynamic indicators, a closed‐loop feedback tuning method was developed to provide patient‐specific EECP treatment strategies. Results showed that for clinical data collected from 113 individuals, the individualized 0D model can accurately simulate patient‐specific global haemodynamic effects (average error <5%). Based on two subjects, the statistical models can be used to evaluate local ATAWSS (error <6%) for coronary arteries and for cerebral arteries. An EECP mode planned by the patient‐specific treatment strategy can promote an appropriate ATAWSS within a 16 s calculation time. The real‐time patient‐specific treatment strategy of EECP is expected to improve the long‐term outcome for each patient and have potential clinical significance. [ABSTRACT FROM AUTHOR]

According to the Universal Design for Learning (UDL) framework, as a textbased alternative to auditory information in videos or presentations, captions can make the content equally accessible, and multilingual subtitles can promote a cross-linguistic understanding of the content. We conducted a phenomenological study to understand the common meaning of the participants' experiences when using real-time automated captions/subtitles during live online class presentations. Twenty-four remote student participants were placed in three study groups. All participants were fluent in spoken and written English, eight could read in one or more additional languages, and none had a hearing disability. We used Microsoft PowerPoint Present Live via Zoom to deliver the online presentation to each group with real-time automated captions/subtitles, and then we conducted a focus group session with each group. Ten themes emerged and were clustered into three overarching themes: challenges, benefits, and interactions with subtitles. Overall, participants described a positive experience, perceiving the captions/subtitles as useful and accurate. Participants found the tool easy to use and highlighted the benefits of using captions/subtitles, such as providing access to live instruction for a wide audience and reinforcing learning for diverse student types. While they were able to troubleshoot connectivity and technological issues encountered, they experienced an apparent split-attention effect and noted limitations in the tool's inability to recognize different dialects. Findings contribute to educational research related to accessible live instruction in multilingual settings and could aid educators in selecting and integrating tools with real-time captioning/subtitling, in line with the UDL guidelines. [ABSTRACT FROM AUTHOR]

Background: Minimally invasive procedures usually require navigating a microcatheter and guidewire through endoluminal structures such as blood vessels and airways to sites of the disease. For numerous clinical applications, two‐dimensional (2D) fluoroscopy is the primary modality used for real‐time image guidance during navigation. However, 2D imaging can pose challenges for navigation in complex structures. Real‐time 3D visualization of devices within the anatomic context could provide considerable benefits for these procedures. Continuous‐sweep limited angle (CLA) fluoroscopy has recently been proposed to provide a compromise between conventional rotational 3D acquisitions and real‐time fluoroscopy. Purpose: The purpose of this work was to develop and evaluate a noniterative 3D device reconstruction approach for CLA fluoroscopy acquisitions, which takes into account endoluminal topology to avoid impossible paths between disconnected branches. Methods: The algorithm relies on a static 3D roadmap (RM) of vessels or airways, which may be generated from conventional cone beam CT (CBCT) acquisitions prior to navigation. The RM is converted to a graph representation describing its topology. During catheter navigation, the device is segmented from the live 2D projection images using a deep learning approach from which the centerlines are extracted. Rays from the focal spot to detector pixels representing 2D device points are identified and intersections with the RM are computed. Based on the RM graph, a subset of line segments is selected as candidates to exclude device paths through disconnected branches of the RM. Depth localization for each point along the device is then performed by finding the point closest to the previous 3D reconstruction along the candidate segments. This process is repeated as the projection angle changes for each CLA image frame. The approach was evaluated in a phantom study in which a catheter and guidewire were navigated along five pathways within a complex vessel phantom. The result was compared to static cCBCT acquisitions of the device in the final position. Results: The average root mean squared 3D distance between CLA reconstruction and reference centerline was 1.87±0.30$1.87 \pm 0.30$ mm. The Euclidean distance at the device tip was 2.92±2.35$2.92 \pm 2.35$ mm. The correct pathway was identified during reconstruction in 100%$100\%$ of frames (n=1475$n=1475$). The percentage of 3D device points reconstructed inside the 3D roadmap was 91.83±2.52%$91.83 \pm 2.52\%$ with an average distance of 0.62±0.30$0.62 \pm 0.30$ mm between the device points outside the roadmap and the nearest point within the roadmap. Conclusions: This study demonstrates the feasibility of reconstructing curvilinear devices such as catheters and guidewires during endoluminal procedures including intravascular and transbronchial interventions using a noniterative reconstruction approach for CLA fluoroscopy. This approach could improve device navigation in cases where the structure of vessels or airways is complex and includes overlapping branches. [ABSTRACT FROM AUTHOR]

Objective: Clinical-standard MRI is the imaging modality of choice for the wrist, yet limited to static evaluation, thereby potentially missing dynamic instability patterns. We aimed to investigate the clinical benefit of (dynamic) real-time MRI, complemented by automatic analysis, in patients with complete or partial scapholunate ligament (SLL) tears. Material and methods: Both wrists of ten patients with unilateral SLL tears (six partial, four complete tears) as diagnosed by clinical-standard MRI were imaged during continuous active radioulnar motion using a 1.5-T MRI scanner in combination with a custom-made motion device. Following automatic segmentation of the wrist, the scapholunate and lunotriquetral joint widths were analyzed across the entire range of motion (ROM). Mixed-effects model analysis of variance (ANOVA) followed by Tukey's posthoc test and two-way ANOVA were used for statistical analysis. Results: With the increasing extent of SLL tear, the scapholunate joint widths in injured wrists were significantly larger over the entire ROM compared to those of the contralateral healthy wrists (p<0.001). Differences between partial and complete tears were most pronounced at 5°–15° ulnar abduction (p<0.001). Motion patterns and trajectories were altered. Complete SLL deficiency resulted in complex alterations of the lunotriquetral joint widths. Conclusion: Real-time MRI may improve the functional diagnosis of SLL insufficiency and aid therapeutic decision-making by revealing dynamic forms of dissociative instability within the proximal carpus. Static MRI best differentiates SLL-injured wrists at 5°–15° of ulnar abduction. [ABSTRACT FROM AUTHOR]

Motorbikes are an integral part of transportation in emerging countries, but unfortunately, motorbike users are also one the most vulnerable road users (VRUs) and are engaged in a large number of yearly accidents. So, motorbike detection is very important for proper traffic surveillance, road safety, and security. Most of the work related to bike detection has been carried out to improve accuracy. If this task is not performed in real-time then it loses practical significance, but little to none has been reported for its real-time implementation. In this work, we have looked at multiple real-time deployable cost-efficient solutions for motorbike detection using various state-of-the-art embedded edge devices. This paper discusses an investigation of a proposed methodology on five different embedded devices that include Jetson Nano, Jetson TX2, Jetson Xavier, Intel Compute Stick, and Coral Dev Board. Running the highly compute-intensive object detection model on edge devices (in real-time) is made possible by optimization. As a result, we have achieved inference rates on different devices that are twice as high as GPUs, with only a marginal drop in accuracy. Secondly, the baseline accuracy of motorbike detection has been improved by developing a custom network based on YoloV5 by introducing sparsity and depth reduction. Dataset augmentation has been applied at both image and object levels to enhance robustness of detection. We have achieved 99% accuracy as compared to the previously reported 97% accuracy, with better FPS. Additionally, we have provided a performance comparison of motorbike detection on the different embedded edge devices, for practical implementation. [ABSTRACT FROM AUTHOR]

Forests are zones of multiple temporalities. They keep time and are constituted through time-keeping practices. Digital technologies of environmental monitoring and management increasingly organise forest temporalities. This article considers how emerging techno-temporalities measure, pace, and transform forest worlds while reproducing and reconfiguring longer durations of colonial and capitalist technologies. We draw together scholarship on political forests, digital media temporalities, and anti-colonial and Indigenous thinking to analyse the politics of time that materialise through digital technologies and shape what forest pasts, presents, and futures are senseable and possible. In particular, we trace the socio-technical production of the 'real-time' as a temporal register of experiencing, knowing, and governing forest environments. Analysing a real-time deforestation alert system in the Amazon, we consider how these temporalities valorise immediate, continuous forest data that can be mobilised for understanding and protecting forests, while simultaneously glossing over durational colonial and capitalist framings of forests that rely on dispossession, extraction, and enclosure. The second half of the article turns to Indigenous futurisms and artistic and socio-political uses of digital platforms that rework forest temporalities. By analysing these multiple and sometimes contradictory temporalities, we suggest that these practices and interventions can challenge dominant timelines and their inequities through pluralistic and redistributive configurations of temporality, land, and data sovereignty. [ABSTRACT FROM AUTHOR]

Thomson scattering of the core edge and divertor plasma regions of a tokamak with reactor technologies is discussed. The rationale and choice of technical solutions are given, the composition of the Thomson scattering diagnostic complex is discussed, as well as an estimate of the accuracy of measuring both electron temperature and density. Particular attention is paid to ensuring the functionality of the proposed diagnostics in the reactor mode of the tokamak operation and the results of testing diagnostic equipment in the experiments on Globus-M2 tokamak. [ABSTRACT FROM AUTHOR]

Purpose: MUC16 is a commonly employed biomarker to identify and predict ovarian cancer (OC). Precise measurement of MUC16 levels is essential for the accurate diagnosis, prediction, and management of OC. This research seeks to introduce a new surface plasmon resonance (SPR) biosensor design that utilizes aptamer-based technology to enable the sensitive and real-time detection of MUC16. Methods: In this study, the sensor chip was immobilized with an anti-MUC16 aptamer (Ap) by utilizing 11-mercaptoundecanoic acid (MUA) as a linker to attach the amine-terminated Ap to the chip using EDC/NHS chemistry. Results: The results indicated that the newly created aptasensor had a detection limit of 0.03 U/mL for MUC16 concentration, with a linear range of 0.09 to 0.27 U/mL. The findings demonstrate good precision and accuracy (<15%) for each MUC16 concentration, with recoveries ranging from 93% to 96%. Additionally, the aptasensor exhibited high selectivity, good repeatability, stability, and applicability in real human serum samples, indicating its potential as a valuable tool for the diagnosis and treatment of OC. Conclusion: According to the outcomes, the designed aptasensor exhibited acceptable specificity to detect the CA125 antigen and could be utilized for the serum detection of target antigen by SPR method. [ABSTRACT FROM AUTHOR]

The term "Innovations in Cyber Security Algorithms for Databases Enhancing Data Retrieval and Management" refers to a book that studies novel techniques for tackling problems related to digital data. The integration of three complicated methods--DQO, DSS, and RAI--is the major focus of attention in this piece of writing. DQO makes use of machine learning to optimize query processing on the fly to meet fluctuating workloads. This is done to accommodate such workloads. To address issues pertaining to the scale of distributed systems, distributed storage systems (DSS) convey data in an effective manner by utilizing consistent hashing. The RAI algorithm adjusts the index architecture in response to the query patterns to achieve real-time flexibility. In this way, the process of looking for information that is frequently asked about is sped up. The methodology that has been suggested is superior to six different ways that are often used in terms of its adaptability, scalability, and real-time capabilities. This article will give a thorough model for improving data management in computer systems. The objective of this essay is to present the model. [ABSTRACT FROM AUTHOR]

Skin cancer is a common and potentially fatal disease that necessitates early and precise diagnosis for effective treatment. In recent years, artificial intelligence has shown promise in aiding dermatologists in the diagnosis of skin cancer. However, the inability of AI models to be interpreted hinders their adoption in clinical practice. This paper presents the design of an AI-based architecture for the real-time diagnosis of skin cancer in an effort to address such issues. The proposed system employs a collection of artificial intelligence (AI) models, including Decision Trees, Rule-Based Models, Logistic Regression, and Deep Forest Models, to achieve accurate and interpretable skin cancer diagnosis. Each model contributes its strengths to the ensemble, thereby enhancing the performance and interpretability of the whole. The ensemble method combines the benefits of various models to compensate for their shortcomings. The effectiveness of the proposed system is demonstrated by the analysis of a Skin Cancer MNIST, ISIC, and Mendeley Skin Cancer Datasets with nearly 250K samples, with 98.9% accuracy, 99.5% precision, and 98.5% recall. The system outperforms existing skin cancer diagnosis methods. The achieved accuracy and performance metrics indicate the system's potential as a reliable real-time diagnostic tool for dermatologists. The proposed system's use cases are diverse. Dermatologists can use the real-time skin cancer diagnosis system to accelerate the screening process, improve diagnostic accuracy, and improve patient outcomes in clinical settings. The models are selected for their ability to capture complex relationships in data, with each model contributing its individual strengths to the ensemble, thereby enhancing the performance and interpretability of the whole. In addition, the system can be integrated into telemedicine platforms, allowing remote patients to receive preliminary assessments and guidance from AI models prior to seeking additional medical care scenarios. [ABSTRACT FROM AUTHOR]

The online detection of small defects within thin steel holds significant importance in the field of steel manufacturing. This paper presents a high-precision flux leakage measurement system based on tunneling magnetoresistance sensors for detecting small defects in thin steel. We optimized the probe size and magnetic circuit design through numerical simulations and used the developed instrument to detect a standard mimicking through-hole defect plate, achieving a high signal-to-noise ratio and successfully detecting artificial defects small to a diameter of 50 μm. Additionally, we detected an actual internal defect in thin steel provided by the steel plant, meeting the required industrial accuracy. The data acquisition process in this study has been optimized to achieve real-time display and defect localization. The measurement scheme proposed in this paper has great potential for online monitoring of the thin steel production process. [ABSTRACT FROM AUTHOR]

Extended Kalman filtering with unknown input (EKF-UI) is often used to estimate the structural system state, parameters and unknown input in structural health monitoring. However, the real-time performance of EKF-UI is bound to whether the measurement equation has a direct feedthrough of unknown input, which great limits its application scope. Based on the zero-order-hold assumption and random walk assumption of unknown input, a novel adaptive discrete state equation is derived in this paper. The new equation establishes a connection between the current state and the current input and allows the adjustment of the sensitivity matrix of the unknown input. Then, based on the adaptive discrete state equation and minimum variance unbiased estimation principle, an adaptive generalized extended Kalman filter with unknown input is derived. The proposed algorithm eliminates the limitation that the real-time performance is restricted by whether the measurement equation has a direct feedthrough of the input and realizes the optimization of the state and input estimates in the sense of minimum variance. To demonstrate the feasibility of the proposed method, numerical example of a shear frame structure with Bouc–Wen hysteresis nonlinearity and experimental test of a five-story shear frame are conducted. The comparison with existing methods shows the advantages of the proposed method. [ABSTRACT FROM AUTHOR]

In this paper, a marker and its real-time tracking method are proposed. Different from existing technologies that recognize the markers by spatial event features, which lead to many practical problems, we discover the possibility of using temporal features. Strobe LEDs (light emitting diode) are used as markers to produce periodically flipping events, and a fast clustering-based algorithm is designed to track and recognize these markers simultaneously. Experiments demonstrate that our methods have superior speed and accuracy compared to state-of-the-arts. The markers can be stably tracked in many challenging situations, thus can be used in various visual tracking applications. The proposed method introduces a new marker and its corresponding recognition algorithm for event camera-based targets tracking, offering a reliable solution for various applications. [ABSTRACT FROM AUTHOR]

In this study, we propose an innovative approach for real-time leakage detection in pipelines by integrating conservative Hamiltonian equations and experimental Internet of Things (IoT) technologies. The proposed method combines a hybrid model that utilizes sensors and IoT devices to acquire real-time data and solves the coupled system of Hamiltonian equations using the ODE45 numerical integration method. Spectral frequency analysis is an essential part of this method, as it reveals specific patterns in the pressure and flow signals. The findings highlighted 95% accuracy in leak detection, which was validated through a comparison of the theoretical and experimental data. The novelty of this approach lies in its ability to maintain constant total system energy, thereby enabling continuous monitoring for early leak detection. As an improvement, the proper handling of sensor signals is emphasized, underscoring its contribution to the efficient management of water resources in potable water distribution systems. [ABSTRACT FROM AUTHOR]

California possesses multiple major freight gateway and logistics facilities that serve both the state and the entire U.S. But the economic, environmental, and local community impacts of trucks, especially heavy-duty trucks that are currently essential to our supply chains and freight transportation system remain poorly measured due to the lack of comprehensive and detailed truck activity data. This paper describes the pilot implementation of the real-time, scalable, and cost-efficient Freight Mobility Living Laboratory (FML2). This system provides truck characterizations across multiple attributes, such as truck body types, axle-based and Gross Vehicle Weight Rating (GVWR)-based classification and is currently deployed at 30 detection locations in Southern California along major freight corridors to support freight modeling and analysis needs. This paper details the design of the FML2 from edge data processing, predictive model development, communication architecture, and backend data storage to the real-time data dashboard to visualize the classification results. Three case studies have been presented at the end of the paper to demonstrate the potential of FML2 for use by both researchers and practitioners to gain further insights on truck activities.

This paper introduces software patterns (registration, acquire-release, and cache awareness) and data structures (Petri net, finite state machine, and protocol flag array) to support the coordinated execution of software activities (also called “components” or “agents”). Moreover, it presents and tests an implementation for Petri nets that supports real-time execution in shared memory for deployment inside one individual robot and separates event firing and handling, enabling distributed deployment between multiple robots. Experimental validation of the introduced patterns and data structures is performed within the context of activities for task execution, control and perception, and decision making for an application on coordinated navigation.

In the modern era, managing optimal real-time control of microgrids during the operation phase has been a significant challenge, requiring careful consideration of both technical and economic factors. This paper introduces a framework for the real-time control of islanded microgrids using a preserving network. This structure incorporates various distributed generation sources, including rotating and non-rotating resources, along with energy storage systems. The optimization function within model predictive control (MPC) manages essential network parameters, such as frequency and voltage, while addressing real-time economic and technical objectives. To enhance precision and account for uncertainties in generation and consumption parameters, the integration of continuous power flow and the preserving network model is employed. This approach aims to create a model that closely mirrors real-world conditions, ensuring a more accurate representation of microgrid dynamics. The proposed structure demonstrates significant improvements in both technical and economic performance compared to Standard MPC and Adaptive MPC, highlighting its potential for more efficient islanded microgrid management. The proposed framework achieves notable reductions in total voltage deviation of 85.87% and 87.62% compared to Standard MPC and Adaptive MPC, respectively. Additionally, it delivers impressive enhancements in frequency deviation of 99.46% and 96.62% compared to Standard MPC and Adaptive MPC, respectively. Economically, the proposed framework significantly outperforms both, reducing costs by 39.29% compared to Standard MPC and by 28.12% compared to Adaptive MPC.

Objective: The metabolism of different cells within the same microenvironment can differ and dictate physiological or pathological adaptions. Current single-cell analysis methods of metabolism are not label-free. Methods: The study introduces a label-free, live-cell analysis method assessing endogenous fluorescence of NAD(P)H and FAD in surface-stained cells by flow cytometry. Results: OxPhos inhibition, mitochondrial uncoupling, glucose exposure, genetic inactivation of glucose uptake and mitochondrial respiration alter the optical redox ratios of FAD and NAD(P)H as measured by flow cytometry. Those alterations correlate strongly with measurements obtained by extracellular flux analysis. Consequently, metabolically distinct live B-cell populations can be resolved, showing that human memory B-cells from peripheral blood exhibit a higher glycolytic flexibility than naïve B cells. Moreover, the comparison of blood-derived B- and T-lymphocytes from healthy donors and rheumatoid arthritis patients unleashes rheumatoid arthritis-associated metabolic traits in human naïve and memory B-lymphocytes. Conclusions: Taken together, these data show that the optical redox ratio can depict metabolic differences in distinct cell populations by flow cytometry.

Economic nonlinear model predictive control is a great choice to tackle the control issues appearing in the current wind energy industry. In this paper, instead of generator power, aerodynamic power is employed in the economic cost function to establish the required conditions of stability and convergence of the economic performance, such as turnpike and problem convexity. However, since aerodynamic power depends on time-varying wind speed, conventional time-invariant economic nonlinear model predictive controllers cannot guarantee the stability and convergence of economic performance. This paper proposes a new time-varying economic nonlinear model predictive controller for wind turbine control that considers an economic trajectory, instead of a steady-state, in its optimization problem. The proposed time-varying economic cost function directly considers aerodynamic power, the activity of pitch angle and generator torque, and fatigue loads on the shaft and tower. Therefore, this controller can maximize power extraction and reduce fatigue load on the tower, drivetrain, and actuators. Furthermore, a fast-parallel Newton-type method is used to implement the proposed controller in actual wind turbines. An accurate aeroelastic model is used to validate the performance of the proposed control scheme. The proposed controller is also compared with two tracking nonlinear model predictive controllers, the baseline controller, and a newly developed method, under fatigue and extreme load scenarios in the presence and absence of uncertainty. The simulation results show the superior economic performance of the proposed approach. Moreover, the real-time results verify the computational speed that the proposed controller requires to deploy actual wind turbines.

This study aims to investigate the differences in intra-urban catchments with different characteristics through real-time wastewater monitoring. Monitoring stations were installed in three neighbourhoods of Barcelona to measure flow, total chemical oxygen demand (COD), pH, conductivity, temperature, and bisulfide (HS-) for 1 year. Typical wastewater profiles were obtained for weekdays, weekends, and holidays in the summer and winter seasons. The results reveal differences in waking up times and evening routines, commuting behaviour during weekends and holidays, and water consumption. The pollutant profiles contribute to a better understanding of pollution generation in households and catchment activities. Flows and COD correlate well at all stations, but there are differences in conductivity and HS- at the station level. The article concludes by discussing the operational experience of the monitoring stations. HIGHLIGHTS Wastewater monitoring stations were deployed in residential catchments.; Flow, chemical oxygen demand, pH, conductivity, temperature, and bisulfide (HS-) were monitored.; Wastewater flow patterns are linked to socioeconomic status.; Catchment characteristics and activities influence loads and concentrations.;

State-of-the-art segmentation models have shown great performance in structured road segmentation. However, these models are not suitable for the wild roads, which are highly unstructured. To tackle the problem of real-time semantic segmentation of wild roads, we propose a Multi-Information Concatenate Network based on BiSeNetV2 and construct a segmentation dataset Dalle Molle institute for artificial intelligence feature segmentation (IDSIAFS) based on Dalle Molle institute for artificial intelligence. The proposed model removes structural redundancy and optimizes the semantic branch based on BiSeNetV2. Moreover, the Dual-Path Semantic Inference Layer (TPSIL) reduces computation by designing the channel dimension of the semantic branch feature map and aggregates feature maps of different depths. Finally, the segmentation results are achieved by fusing both shallow detail information and deep semantic information. Experiments on the IDSIAFS dataset demonstrate that our proposed model achieves an 89.5% Intersection over Union. The comparative experiments on Cityscapes and India driving dataset benchmarks show that proposed model achieves good inference accuracy and faster inference speed.

Presently, e-commerce platforms incorporate image search functionalities. Nevertheless, these systems possess constraints; input images necessitate static and manual cropping since the system does not automatically generate bounding boxes. Addressing this concern requires the implementation of an object detection algorithm to ascertain the quantity, location, and type of desired objects within real-time bounding boxes before users finalize their selection. This capability empowers users to readily discern their desired items, thereby augmenting the precision and efficiency of visual searches. Despite the availability of swifter object detection algorithms such as R-CNN and Mask R-CNN, which prioritize accuracy over speed, rendering them less suited for real-time detection, we opted to employ the YOLOv4 algorithm as an alternative, renowned for its efficacy in real-time object detection. Furthermore, we adopted the Color, Texture, and Edge-Based Image Retrieval (CTEBIR) technique for image matching. The results of our experimentation demonstrate that the utilization of the YOLOv4 algorithm can enhance the accuracy and speed of visual searches by streamlining the search process based on the identified classes. Additionally, our precision assessment yielded a score of 95%, with individual scores for camera objects reaching 90%, keyboards achieving 85%, and laptops attaining 71%. These findings corroborate the dependability of the CTEBIR algorithm in image matching and contribute to a deeper comprehension of the system's efficacy in accurately detecting and distinguishing objects.

Introduction: Despite the many attempts made by researchers to diagnose jaundice non-invasively using machine learning techniques, the low amount of data used to build their models remains the key factor limiting the performance of their models. Objective: To build a system to diagnose neonatal jaundice non-invasively based on machine learning algorithms created based on a dataset comprising 767 infant images using a computer device and a USB webcam. Methods: The first stage of the proposed system was to evaluate the performance of four machine learning algorithms, namely support vector machine (SVM), k nearest neighbor (k-NN), random forest (RF), and extreme gradient boost (XGBoost), based on a dataset of 767 infant images. The algorithm with the best performance was chosen as the classifying algorithm in the developed application. The second stage included designing an application that enables the user to perform jaundice detection for a patient under test with the minimum effort required by capturing the patient’s image using a USB webcam. Results: The obtained results of the first stage of the machine learning algorithms evaluation process indicated that XGBoost outperformed the rest of the algorithms by obtaining an accuracy of 99.63%. The second-best algorithm was the RF algorithm, which had an accuracy of 98.99%. Following RF, with a slight difference, was the k-NN algorithm. It achieved an accuracy of 98.25%. SVM scored the lowest performance among the above three algorithms, with an accuracy of 96.22%. Based on these obtained results, the XGBoost algorithm was chosen to be the classifier of the proposed system. In the second stage, the jaundice application was designed based on the model created by the XGBoost algorithm. This application ensured it was user friendly with as fast a processing time as possible. Conclusion: Early detection of neonatal jaundice is crucial due to the severity of its complications. A non-invasive system using a USB webcam and an XGBoost machine learning technique was proposed. The XGBoost algorithm achieved 99.63% accuracy and successfully diagnosed 10 out of 10 NICU infants with very little processing time. This denotes the efficiency of machine learning algorithms in healthcare in general and in monitoring systems specifically.

The most important information required to successfully issue a flood warning is the quantitative precipitation forecasts (QPFs). This is important to run subsequent rainfall–runoff simulations. A rainfall–runoff simulation derives its accuracy mainly from the accuracy of the input QPFs. The dynamically based global numerical weather prediction models (NWPMs) are strong candidate sources of QPFs. A main problem is the real-time selection of which NWPM should be used to provide the QPFs for flood warning simulations. This paper develops an automated technique to solve this problem. The technique performs real-time comparisons with measured rainfall fields using a novel ‘tolerant’ hydrologic approach. The ‘tolerant’ approach performs the comparison on the basin scale and allows for timing shifts in the forecasts. This is because QPFs can be good but only a few hours early or late. Two events are used for illustration, and the proposed real-time application in flood warning is presented. The developed technique, employing the tolerant approach, could eliminate the effects of the timing shifts and, accordingly, succeeded to select the QPFs to be used. A Python package was developed for automation. The developed technique is expected to also be useful for offline assessments of historical performances of NWPMs. HIGHLIGHTS An automated technique was developed for the assessment of numerical weather prediction model (NWPM) forecasts.; This technique employs a novel tolerant hydrologic approach imitating the human eye/brain judgment for evaluating the forecasting skills.; Real-time selection of the best NWPM, including real-time bias and uncertainty estimation, for flood warning applications is described.; A Python package for automation was developed.;

The push to reduce size, weight, power, and cost (SWaP-C) has led to the integration of multiple subsystems with different criticality levels onto a single hardware platform. Virtualization has played a crucial role in merging real-time capabilities with complex environments, facilitating the creation of Mixed-Criticality Systems (MCS). However, consolidating various workloads on a single platform often requires shared peripheral access, such as Network Interface Controllers (NICs), among guest Virtual Machines (VMs). VirtIO provides a standardized interface for device para-virtualization, enabling us to deploy device drivers and sharing logic within dedicated VMs. Despite its advantages, VirtIO can introduce latency and reduced throughput due to its reliance on the hypervisor or other VMs for network access, which may exacerbate resource contention and affect time isolation between VMs. Time-Sensitive Networking (TSN) techniques can mitigate these issues by ensuring reliable, low-latency, and deterministic network communication, although its implementation often depends on proprietary Linux-based solutions. In this paper, we present BiRtIO, an infrastructure designed to support a virtual real-time network for VMs on the Bao hypervisor without compromising safety and security. It leverages the VirtIO standard and incorporates TSN mechanisms available in mainline Linux, such as bandwidth limitation, to enhance performance and predictability in a saturated virtualized network. Experiments conducted on the Xilinx ZCU102 high-performance multicore platform demonstrate that our solution achieves performance and predictability levels suitable for real-time workloads on MCS with limited Ethernet ports.

Food cooking process are complex dynamical systems to model. In the state of the art we find that a good solution consists of physics-based finite element models (FEM). FEM models, although being very accurate, have a high computational cost making them unfeasible for real-time applications. To solve this problem, we consider neural networks (NN) trained from FEM simulations. Specifically, we propose a Nonlinear AutoRegressive with eXogenous inputs Neural Network (NARX-NN). The main novelty is that we define a novel training algorithm adapted to the modeling of real-time dynamical systems, allowing a NARX-NN with a simple structure to obtain a negligible error compared to the results of the original FEM model. The NARX-NN trained with the proposed training algorithm obtains an R-squared of more than 0.99 in the test simulations, while the same NARX-NN trained with the standard training algorithm obtains an R-squared of 0.78 in the same tests. The proposed NARX-NN achieves a speedup of 8 orders of magnitude compared to the original FEM model. Moreover, the developed NN is able to predict the complete cooking of the food in a few milliseconds without the need of external sensors. Alternatively, our approach can also be used in real time with information captured with sensors. The presented methodology is highly scalable and could be adapted to different types of food and cooking processes, as well as to other dynamical systems in general.

The ability to simulate realistic crowds is a highly sought-after capability in the fields of entertainment (video games, movies), urban planning and evacuation simulations. Traditional approaches to crowd simulation rely on heavy Central Processing Unit (CPU) computation. This approach has limitations in terms of scalability and performance, which are solvable with the use of Graphics Programming Units (GPUs) and parallel computing techniques. In fact, the development of Compute Shaders on GPU allows the execution of general-purpose operations alongside traditional rendering tasks within real-time applications. This paper aims to contribute to the current literature on crowd simulation methods by developing a real-time simulation model that integrates and expands several techniques from literature, adapted and optimized to exploit GPU computing capabilities. The proposed model incorporates continuous representations for crowds in order to simulate human movement and decision-making. The achieved results demonstrate a high level of scalability and efficiency. The implemented techniques and optimizations allow the model to handle a significant number of agents while maintaining real-time performances to achieve reduced simulation time and good user experience. Stress tests showcase that the proposed model significantly outperforms other macroscopic models, maintaining a stable frame rate of 60 FPS when simulating 20,000 agents even on mid-range systems intended for personal use.

The rise of edge computing has introduced unique challenges for deploying efficient AI solutions in resource-limited environments. While traditional AI frameworks are powerful, they often fall short in meeting the requirements of edge computing, such as low latency, constrained computational power, and energy efficiency. This paper presents Polara-Keras2c, an optimized evolution of Keras2c designed specifically for edge computing. Polara-Keras2c enhances compatibility with bare-metal systems, incorporates RISC-V vector extension optimization, and is customized for the Polara architecture. By converting pre-trained Keras models into optimized C code for bare-metal execution on edge devices, Polara-Keras2c enables advanced AI models to operate efficiently in resource-constrained environments. The framework supports fixed-point arithmetic, achieving a minimal accuracy impact of only 0.03% when tested on the MNIST dataset, and offers a streamlined approach for rapid prototyping. Experimental results reveal that Polara-Keras2c achieves up to 4.81 times faster convolution processing with a $64\times 64$ input size compared to scalar processing, significantly enhancing computational efficiency and reducing energy consumption. These capabilities position Polara-Keras2c as a transformative tool in real-time, energy-efficient AI processing for edge devices, pushing forward the evolution of edge computing.

Convolutional Neural Networks (CNNs) have been utilised in many image and video processing applications. The convolution operator, also known as a spatial filter, is usually a linear operation, but this linearity compromises essential features and details inherent in the non-linearity present in many applications. However, due to its slow processing, the use of a nonlinear spatial filter is a significant bottleneck in many software applications. Further, due to their complexity, they are difficult to accelerate in FPGA or VLSI architectures. This paper presents novel FPGA implementations of linear and nonlinear spatial filters. More specifically, the arithmetic computations are carried out in custom floating-point, enabling a tradeoff of precision and hardware compactness, reducing algorithm development time. Further, we show that it is possible to process video at a resolution of 1080p with a frame rate of 60 frames per second, using a low-cost FPGA board. Finally, we show that using a domain-specific language will allow the rapid prototyping of image processing algorithms in custom floating-point arithmetic, allowing non-experts to quickly develop real-time video processing applications.

The most advanced DC shipboard power systems will be based on the wide deployment of controlled converters in a zonal distribution configuration. On one side the power electronics arrangement enables advantages for the onboard functioning (e.g. flexibility, redundancy and adaptability), on the other it can open worrying issues on DC system stability. When perturbing a negative-damped DC power grid, the incoming instability triggers a ship blackout as a consequence. To avoid this catastrophic eventuality, the Power Management System (PMS) is programmed for the smart real-time adaptation of converters bandwidths. The paper presents a Preventive-Resilient (P-R) bandwidths tuning to guarantee the stable functioning, even when assuming the (n−1) eventuality, so the unexpected disconnection of the more stabilizing load. This reconfiguration is an effective measure to constantly ensure a sufficient stability margin, even when low-bandwidth loads are offline. The capability of P-R algorithm is evaluated by a C-HIL test on a digital real-time platform. A Python code acts as a PMS in configuring the bandwidths on emulated converters via the Modbus communication protocol. The combination of a simplified stability assessment method with an optimization algorithm enables a fast bandwidth reconfiguration process without requiring excessive computational effort. These characteristics make the proposed approach suitable for integration into the shipboard Power Management System (PMS), providing central controllers with an online stability assessment and safe load management tool.

Infrared template matching is an essential technology that enables reliable and accurate object detection, recognition, and tracking in complex environments. Perceptible Lightweight Zero-mean normalized cross-correlation (ZNCC) Template Matching (PLZ-TM) has been proposed as a tool for matching infrared images obtained from cameras with different fields of view. Aligning such images is challenging because of the involved differences in thermal distributions, focus discrepancies, background elements, and distortions. The first stage of PLZ-TM involves extracting feature maps from the search and template images using a deep learning network. This deep learning network is designed with a Convolutional Neural Network (CNN) architecture that omits pooling layers, thereby minimizing information loss during extraction. The subsequent stage involves matching the feature maps. The matching method utilizes a lightweight ZNCC (ZNCC) module that employs average pooling for training. The deep learning network is trained to optimize the distribution of the output heatmap and the probability at the correct location of the template image. PLZ-TM delivers excellent performance achieving a processing time of only 3.3 ms in matching a $640\times 480$ search image with a $192\times 144$ template image. Moreover, it attains a matching accuracy of 96% on a dataset obtained from infrared cameras with different fields of view.

Stereo vision technology, as a significant branch of computer vision, has been widely applied in fields such as robot navigation, autonomous driving, and 3D reconstruction. Achieving embodied intelligence through edge hardware platforms remains challenging in balancing power consumption, real-time performance, and accuracy. Although the semi-global stereo matching algorithm has proven effective in balancing disparity map accuracy and computational complexity, its matching accuracy is often limited by weak textures, disparity discontinuities, lighting variations, and noise. To address the limitations of existing semi-global stereo matching algorithms and the power constraints of hardware platforms, this paper proposes a stereo matching algorithm based on FPGA that integrates the Sum of Absolute Differences (SAD) with adaptive Census transform. The goal is to enhance image and edge information and design a compact and efficient stereo vision hardware accelerator. During the hardware implementation phase, a pixel-level pipelined parallel matching cost computation structure is proposed. This structure significantly reduces data buffer requirements through multi-step parallel computation. Additionally, a two-stage four-layer parallel pipelined semi-global cost aggregation architecture is adopted, which effectively balances hardware resource utilization while maintaining accuracy. Evaluations of the Middlebury dataset show that compared to traditional SAD and Census algorithms, the proposed algorithm improves matching accuracy by 15.67% and 15.1%, respectively. On the Xilinx Zynq-7 platform, for images with resolutions of $1280\times 720$ and $640\times 480$ , the processing speeds reach 54.24fps and 81.34fps, respectively.

Real-time RGB-D semantic segmentation is crucial for tasks like dynamic environment analysis of mobile robots or autonomous vehicles, real-time and efficient segmentation results can enhance the accuracy of subsequent tasks such as free space detection, mapping and navigation. It requires models to balance computational cost and performance by employing more efficient mechanisms to effectively recognize differences in RGB-D multimodal information, retain complementary information, and reduce redundancies. Currently, although convolutional neural network (CNN) methods are less accurate than Transformer-based methods, they offer stronger real-time performance under the same computational load. Therefore, in this study, we proposed the Efficient Feature Interaction Network (EFINet), a real-time RGB-D segmentation method that uses a lightweight CNN encoder and incorporates encoder blocks with a lightweight upsampling method Dysample and the carefully optimized number of ConvNeXt V2 blocks, to redesign the decoder and minimize redundant computations. Additionally, we propose a robust cross-modal interaction mechanism that facilitates the exchange of useful information and the elimination of noise between modalities with minimal computational overhead. By incorporating lightweight convolutions, the proposed method can acquire multi-scale global information with minimal additional computation, thus providing the model with rich contextual information. The proposed EFINet has 38.0 M parameters and 40.0 GFlops, achieving 55.5% and 50.6% mIoU on the challenging NYUDv2 and SUN RGB-D datasets, respectively, and real-time inference at 31.2 FPS. EFINet outperforms several state-of-the-art methods in the real-time RGB-D segmentation task while achieving a better balance between segmentation performance and computational efficiency.

In recent years, there has been an increasing need for effective voltage control methods in power systems due to the growing complexity and dynamic nature of practical power grid operations. To enhance the controller’s resilience in addressing communication failures, a dynamic voltage control method employing distributed execution multi-agent deep reinforcement learning(DRL) is proposed. The proposed method follows a centralized training and decentralized execution based approach. Each agent has independent actor neural networks to output generator control commands and critic neural networks that evaluate command performance. Detailed dynamic models are integrated for agent training to effectively capture the system’s dynamic behavior following disturbances. Subsequent to training, each agent possesses the capability to autonomously generate control commands utilizing only local information. Simulation outcomes underscore the efficacy of the distributed execution multi-agent DRL controller, showcasing its capability in not only providing voltage support but also effectively handling communication failures among agents.