Real-time Generation Research Articles

Hand Gesture Recognition for Interactive On-Screen Drawing with Real-Time Subtitling for Teaching

This research presents the development of an innovative educational tool leveraging hand gesture recognition for interactive on-screen drawing with real-time subtitling. Built on OpenCV and Python, the platform enables educators to draw and manipulate digital content through intuitive hand gestures, enhancing classroom engagement. Additionally, real-time subtitle generation offers accessibility for hearing-impaired students by converting drawn content into text notes. This system addresses key challenges in traditional digital drawing tools, such as limited interaction methods and accessibility, aiming to deliver a more inclusive learning environment. Preliminary results indicate high recognition accuracy, smooth interaction, and potential for broader educational applications. Future improvements will involve expanding the gesture set and refining accuracy for complex gestures. Keywords: Hand Gesture Recognition, Interactive Learning, Real-Time Subtitling, Educational Technology, OpenCV

A Novel BEM for the Hydrodynamic Analysis of Twin-Hull Vessels with Application to Solar Ships

A novel Boundary Element Method (BEM) is presented for predicting the hydrodynamic behavior of twin-hull vessels, traveling at low speeds, aiming to quantify the benefits of integrating solar technology onboard. In particular, the power requirements of an electric 33 m long twin-hull ship are examined. The study discusses the placement of solar panels on deck and assesses their utilization in terms of real-time energy generation, aiming to extend the autonomy range while also reducing carbon emissions. The discussed methodology predicts the power needs by considering various operational variables, design specifications and hydrodynamic principles. In addition, it addresses the viability and possible advantages of integrating solar technology onboard and provides preliminary estimates regarding the extent to which solar energy may compensate for power needs, based on several factors, including the velocity, the prevailing sea state and the incident solar irradiance. The results provide useful information regarding the utilization of solar energy in the shipping sector, in addition to advancing sustainable maritime propulsion.

Low-Cost Constant Time Signed Digit Selection for Most Significant Bit First Multiplication

Serial binary multiplication is frequently used in many digital applications. In particular, left-to-right (aka online) manipulation of operands promotes the real-time generation of product digits for immediate utilization in subsequent online computations (e.g., successive layers of a neural network). In the left-to-right arithmetic operations, where a residual is maintained for digit selection, utilization of a redundant number system for the representation of outputs is mandatory, while the input operands and the residual may be redundant or non-redundant. However, when the input data paths are narrow (e.g., eight bits as in BFloat16), conventional non-redundant representations of inputs and residual provide some advantages. For example, the immediate and costless sign detection of the residual that is necessary for the next digit selection; a property not shared by redundant numbers. Nevertheless, digit selection, as practiced in the previous realizations, with both redundant and non-redundant inputs and/or residual, is slow and rather complex. Therefore, in this paper, we offer an imprecise, but faster digit selection scheme, with the required correction in the next cycle. Analytical evaluations and synthesis of the proposed circuits on FPGA platform, shows 30% speedup and less cost with respect to both cases with redundant and non-redundant inputs and residual.

AI-Driven Traffic Simulation using Unity: Implementing Finite State Machines for Adaptive NPC Behaviour

This research develops an AI-powered traffic simulation using the Unity Engine, leveraging finite state machines (FSM) to enable adaptive and responsive non-player characters (NPCs). The integration of FSM with advanced pathfinding algorithms, such as A*, allows NPCs to dynamically adjust their behavior based on traffic conditions, obstacles, and environmental changes. The experimental results indicate a 25% improvement in route optimization and a 30% reduction in path conflicts compared to conventional static models, demonstrating the robustness of the proposed approach. Optimized navmesh deployment further enhances navigation fluidity, ensuring efficient agent movement in high-density scenarios without compromising system performance. The findings establish the effectiveness of the FSM-driven NPC behavior in simulating realistic traffic environments, contributing both to the advancement of AI applications in game development and urban planning. By providing an interactive platform for traffic management, this simulation offers a practical tool to study congestion patterns and test intervention strategies. In addition, it improves player engagement by fostering emergent gameplay through dynamic NPC interactions. Future work could explore the integration of real-time procedural generation or multiplayer functionality to enrich simulation depth and scalability. This study provides a comprehensive framework that bridges AI-based mechanics with simulation technology, providing significant insights for researchers and practitioners in game design, artificial intelligence, and urban planning.

Utilizing ChatGPT for On-the-Fly Decision-Making in Autonomous Systems

In the realm of autonomous systems, rapid and efficient decision-making is critical to ensure optimal performance and adaptability in dynamic environments. This paper explores the application of ChatGPT, a large-scale language model, in on-the-fly decision-making within autonomous systems. By leveraging ChatGPT's capabilities in natural language processing, contextual understanding, and real-time response generation, the study demonstrates how it can support autonomous agents in interpreting situations, generating actionable insights, and making informed decisions. The integration of ChatGPT into autonomous systems provides a flexible and scalable approach to improving decision accuracy and response times in unpredictable scenarios. This research investigates the benefits, challenges, and potential use cases, highlighting the model's contributions to enhancing autonomy in various fields such as robotics, self-driving vehicles, and intelligent drones.

Innovating Statistics Education: The Design of a Novel App Using Design Thinking

Statistical education at university level faces significant challenges, particularly with the rapid advancements in technology and evolving teaching methods. Updating teaching methodologies for statistics in higher education is essential. Information technologies can greatly enhance the learning of statistical content, but many existing platforms are not well-suited for university contexts, either due to language barriers or how the content is presented (often confusing, with limited explanations or lacking context). It is crucial to have a student-centered platform that addresses these issues, ensuring an effective and efficient learning experience. This study introduces “EstApp”, an innovative mobile application prototype designed for teaching and learning descriptive statistics in university courses. The application was developed using the Design Thinking methodology, which emphasizes user experience and focuses on the needs of the end users. The design process involved stages of empathizing, defining, ideating, prototyping, and testing, culminating in validation through user tests with university students and professors. This study concludes that EstApp’s functionalities enhance understanding of (1) statistical models through interactive graphs and data visualizers; (2) probability concepts via a probability calculator; and (3) descriptive statistics through real-time data generation and visualizers.

Generative domain-adapted adversarial auto-encoder model for enhanced ultrasonic imaging applications

In this study, we propose a class-conditioned Generative Adversarial Autoencoder (cGAAE) to improve the realism of simulated ultrasonic imaging techniques, in particular the Multi-modal Total Focusing Method (M-TFM), based on the availability of both simulated and experimental TFM images. In particular, this work studied the case of the inspection of a complex geometry block representative of weld-inspection problem based on ultrasonic multi-elements probe. The cGAAE is represented by a tailored learning schema, trained in a semi-supervised fashion on a labeled mixture of synthetic (class 0) and experimental (class 1) M-TFM images, obtained under different meaningful inspection set-ups parameters (i.e., the celerity of the transverse ultrasonic wave, the specimen back-wall slope and height, the flaw tilt and heigh). That is, the cGAAE schema consists in a combination of learning stages involving class-conditioned spatial-transformers and arbitrary style transfer endows the cGAAE of powerful generative features, such as quasi real-time generation of M-TFM images by sweep of the inspection parameters. We exploited the cGAAE model to improve the realism of simulated M-TFM images and enhance the accuracy of the inverse problem, aiming at estimating the inspection parameters based on experimental acquisitions.

Bio-Inspired Motion Emulation for Social Robots: A Real-Time Trajectory Generation and Control Approach.

Assistive robotic platforms have recently gained popularity in various healthcare applications, and their use has expanded to social settings such as education, tourism, and manufacturing. These social robots, often in the form of bio-inspired humanoid systems, provide significant psychological and physiological benefits through one-on-one interactions. To optimize the interaction between social robotic platforms and humans, it is crucial for these robots to identify and mimic human motions in real time. This research presents a motion prediction model developed using convolutional neural networks (CNNs) to efficiently determine the type of motions at the initial state. Once identified, the corresponding reactions of the robots are executed by moving their joints along specific trajectories derived through temporal alignment and stored in a pre-selected motion library. In this study, we developed a multi-axial robotic arm integrated with a motion identification model to interact with humans by emulating their movements. The robotic arm follows pre-selected trajectories for corresponding interactions, which are generated based on identified human motions. To address the nonlinearities and cross-coupled dynamics of the robotic system, we applied a control strategy for precise motion tracking. This integrated system ensures that the robotic arm can achieve adequate controlled outcomes, thus validating the feasibility of such an interactive robotic system in providing effective bio-inspired motion emulation.

Abstract C008: Patient-derived organoids and precision medicine: Insights from the PASS-01 clinical trial in PDAC

Abstract Introduction: Pancreatic ductal adenocarcinoma (PDAC) is challenging as most patients are diagnosed at an advanced stage with limited treatment options. Tailoring therapies for individual patients is paramount due to the aggressive nature of the disease. Our work explores patient- derived organoids (PDOs) as in vitro tumor models to dissect molecular signatures and conduct pharmacotyping, aiming to enhance precision medicine. Methods: In the PASS-01 stage IV PDAC clinical trial, biopsies were collected for molecular correlatives, including establishment of PDO models. Tissue samples were collected from patients at six institutes across the United States and Canada, highlighting our ability to integrate PDO models into a robust, multi- institutional clinical framework. Most biopsies were collected from liver metastases, followed by primary pancreas tumors, peritoneal, omental, lymph node, lung, and brain metastases. KRAS mutation status by ddPCR was used to validate neoplastic cells in the organoid cultures and identify pseudonormal outgrowth. Established PDO lines were subjected to high throughput drug screening for 120+ compounds, comprising both standard-of-care (as monotherapy and in combination therapies) and experimental agents. Subsequently, validated PDO lines were expanded, biobanked and harvested for RNA and DNA sequencing. Results: During the trial, 186 biopsies from 183 enrolled patients were shipped to CSHL for PDO establishment. The overall malignant PDO establishment rate was 50% across all biopsy sites and patients. Interestingly, PDOs could be generated more often from patients that rapidly progressed on therapy (p = 0.03). Furthermore, by comparing the characteristics of the primary biopsies to the established PDO, PDOs were more frequently generated from Moffitt subtype classical (64% establishment) compared to basal (35%), and from those with a KRASminor imbalance (67% establishment) compared with KRAS wild-type (36%), balanced (58%), and KRASmajor imbalance (52%). The average time from tissue receipt to first drug screen data was 65 days, with six PDOs screened in under 30 days. This turnaround time enabled PDO therapeutic data to be presented at monthly molecular tumor boards. Consequently, these data were considered alongside other clinical trial correlates to aid in selection of second-line therapies when patients progressed. Conclusions: The PASS-01 trial facilitated the real-time generation of PDO models and reporting of therapeutic results. We identified correlations between PDO establishment and patient characteristics, and improved the methods to detect pseudonormal outgrowth. PDO pharmacoptyping identified sensitivity that correlated with patient outcomes on GnP, while also highlighting challenges of using empiric drug testing for chemotherapy sensitivity. Moving forward we aim to continue incorporation of PDOs and tumor molecular profiling to aid in patient therapy selection. We anticipate this work to be crucial as targeted therapies, including KRAS inhibitors, become mainstream in PDAC care. Citation Format: Amber N. Habowski, Dennis Plenker, Hardik Patel, Caitlin Tsang, Luce St. Surin, Fatim Kouassi, Deepthi Budagavi, Grainne M O'Kane, Stephanie Ramotar, Kenneth H Yu, Faiyaz Notta, Andrew Aguirre, Brian Wolpin, Dan Laheru, Daniel A King, Elizabeth M Jaffee, Jennifer J Knox, David A Tuveson. Patient-derived organoids and precision medicine: Insights from the PASS-01 clinical trial in PDAC [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Advances in Pancreatic Cancer Research; 2024 Sep 15-18; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2024;84(17 Suppl_2):Abstract nr C008.

Real-time solar PV generation in a building using LSTM-based time series forecasting

The increasing demand for renewable energy has forced the researchers to lay importance on studying the behavior of these sources by monitoring the operational data and performing data analysis. The application of data analytics to renewable energy is one of the most powerful methods to maintain the system's reliability and stability. As the power system becomes more complex the necessity to perform data analytics also increases. When data analytic techniques are applied to solar energy generations through Photovoltaic (PV) dataset, the possible behavior of PV generation performance which is affected by changes in environmental conditions can be predicted and further analytical approaches allow us to detect possible PV panel and inverter failures. This paper is an attempt towards applying the intelligent data analytics approaches to solar PV generation of a real-time photovoltaic plant. The main purpose of the data analytics platform is to analyze the energy yield, assess the PV system performance, and forecast solar PV generation. In this article, Long Short-Term Memory (LSTM) machine learning model is developed to assess and interpret the available information from the gathered data of the PV plant. The proposed model is a sequence-to-sequence regression-based forecasting model which performs time series forecasting in a real-time 100 kW PV plant in the University building of SRM Institute of Science and Technology, Kattankulathur, Tamilnadu.

Dynamic Scheduling Optimization of Automatic Guide Vehicle for Terminal Delivery under Uncertain Conditions

As an important part of urban terminal delivery, automated guided vehicles (AGVs) have been widely used in the field of takeout delivery. Due to the real-time generation of takeout orders, the delivery system is required to be extremely dynamic, so the AGV needs to be dynamically scheduled. At the same time, the uncertainty in the delivery process (such as the meal preparation time) further increases the complexity and difficulty of AGV scheduling. Considering the influence of these two factors, the method of embedding a stochastic programming model into a rolling mechanism is adopted to optimize the AGV delivery routing. Specifically, to handle real-time orders under dynamic demand, an optimization mechanism based on a rolling scheduling framework is proposed, which allows the AGV’s route to be continuously updated. Unlike most VRP models, an open chain structure is used to describe the dynamic delivery path of AGVs. In order to deal with the impact of uncertain meal preparation time on route planning, a stochastic programming model is formulated with the purpose of minimizing the expected order timeout rate and the total customer waiting time. In addition, an effective path merging strategy and after-effects strategy are also considered in the model. In order to solve the proposed mathematical programming model, a multi-objective optimization algorithm based on a NSGA-III framework is developed. Finally, a series of experimental results demonstrate the effectiveness and superiority of the proposed model and algorithm.

Synthesis and characterization of amine-functionalized graphene as a nitric oxide-generating coating for vascular stents.

Drug-eluting stents are commonly utilized for the treatment of coronary artery disease, where they maintain vessel patency and prevent restenosis. However, problems with prolonged vascular healing, late thrombosis, and neoatherosclerosis persist; these could potentially be addressed via the local generation of nitric oxide (NO) from endogenous substrates. Herein, we develop amine-functionalized graphene as a NO-generating coating on polylactic acid (PLA)-based bioresorbable stent materials. A novel catalyst was synthesized consisting of polyethyleneimine and polyethylene glycol bonded to graphene oxide (PEI-PEG@GO), with physicochemical characterization using x-ray diffraction, Raman spectroscopy, Fourier transform infrared spectroscopy, and thermogravimetric analysis. In the presence of 10 μM S-nitrosoglutathione (GSNO) or S-nitroso-N-acetylpenicillamine (SNAP), PEI-PEG@GO catalyzed the generation of 62% and 91% of the available NO, respectively. Furthermore, PEI-PEG@GO enhanced and prolonged real-time NO generation from GSNO and SNAP under physiological conditions. The uniform coating of PEI-PEG@GO onto stent material is demonstrated via an optimized simple dip-coating method. The coated PLA maintains good biodegradability under accelerated degradation testing, while the PEI-PEG@GO coating remains largely intact. Finally, the stability of the coating was demonstrated at room temperature over 60 days. In conclusion, the innovative conjugation of polymeric amines with graphene can catalyze the generation of NO from S-nitrosothiols at physiologically relevant concentrations. This approach paves the way for the development of controlled NO-generating coatings on bioresorbable stents in order to improve outcomes in coronary artery disease.

Energy-efficient real-time visual image adversarial generation and processing algorithm for new energy vehicles

Energy-efficient real-time visual image adversarial generation and processing algorithm for new energy vehicles

Developing a digital data platform for surveillance of food and water-borne pathogens in North East India: insight for public health advocacy.

Robust digital infrastructure is vital and the need of the hour, especially in the healthcare sector, for real-time data generation, analysis, and quick decision-making. Food- and water-borne illnesses represent a prominent cause of morbidity and mortality worldwide. India, a developing nation with diverse cultures and food practices, poses a high risk of food-borne diseases and outbreaks, yet is often underreported and ineffectively researched. Also, the unique socio-economic and environmental factors of the Northeast (NE) region contribute to the high burden of food-borne diseases. To address these trepidations, the Indian Council of Medical Research (ICMR) has undertaken a study for the surveillance of food-borne pathogens in NE India. The present study focuses on the development of a digital database system for the systematic surveillance of foodborne disease outbreaks, aiming to address the gaps in traditional surveillance methods and improve disease detection and response capabilities. The digital system integrates mobile applications, web-based platforms, and advanced analytics tools to enable real-time data collection, dissemination, and analysis of food-borne illness data. Additionally, the secure and scalable nature of the system enhances data accuracy and accessibility, making it a valuable tool for enhancing food-borne disease surveillance efforts in resource-constrained settings.

VAEnvGen: A Real-Time Virtual Agent Environment Generation System Based on Large Language Models

Environment plays an important role in non-verbal communication for human-virtual agent interaction. Existing research explores the influence of an agent’s appearance and attributes to enhance human-virtual agent communication. However, there is no common practice for dynamically adjusting the surrounding environments of the virtual agent. In this paper, we introduce a real-time virtual agent environment generation system (VAEnvGen), which contributes to the field by enhancing users’ content perception and improving task performance through dynamic environment adjustment. The system dynamically analyzes both the appropriate communication environment and filters the key information according to the current context. Leveraging Large Language Models, it generates a pseudo-3D background space to create an engaging atmosphere and a dynamic foreground content space for vivid key information display, thereby significantly enhancing content perception. For widespread adoption and flexibility, VAEnvGen is developed as a web application. We further evaluate the impact of VAEnvGen on content perception, user attention, and subjective satisfaction through a mixed-design user study with 50 participants. Quantitative and qualitative results reveal significant improvements in content perception, task completion time, and user satisfaction when using VAEnvGen. The system effectively redistributes user attention from subtitles and the virtual agent itself to the dynamically generated background and key foreground information, leading to a more immersive and less fatiguing user experience.

Real-Time Procedural Generation with GPU Work Graphs

We present a system for real-time procedural generation that makes use of the novel GPU programming model, work graphs. The nodes of a work graph are shaders, which dynamically generate new workloads for connected nodes. This greatly simplifies the implementation of recursive procedural algorithms on GPUs. Combined with GPU ray tracing and procedural mesh shaders, our system makes use of this graph structure to tackle various common problems of procedural generation. Our system is very easy to implement, requiring no additional data structures from what would already be available in a modern rendering engine. We demonstrate the real-time editing capabilities on representative examples. We augment the scene in the teaser image with 79,710 instances in 3.74 ms on an AMD Radeon RX 7900 XTX.

Trading electricity using tranched power delivery contracts

ABSTRACT This letter proposes a novel way to apportion real-time renewable generation outputs between diverse token holders, using a novel tranching mechanism. This novel tranching structure is proposed to better manage the unpredictability of renewable power delivery and also as a case study of the exotic new instruments that could be offered in decentralized electricity marketplaces. Under this framework, a wind farm directly sells claims on their future power output in the form of a digital token. The realized generation then offsets the token holders’ electricity consumption in near real-time. Two ways of structuring such forward delivery instruments are considered: a tranched system, where more senior token holders enjoy priority claims on available outputs, as compared against a pro-rata scheme, where the realized generation is equally apportioned between token holders. A notional market simulation is provided to explore how consumers may exploit the distinct power delivery profiles.

A Secure Authentication Scheme with Local Differential Privacy in Edge Intelligence-Enabled VANET

Edge intelligence is a technology that integrates edge computing and artificial intelligence to achieve real-time and localized model generation. Thus, users can receive more precise and personalized services in vehicular ad hoc networks (VANETs) using edge intelligence. However, privacy and security challenges still exist, because sensitive data of the vehicle user is necessary for generating a high-accuracy AI model. In this paper, we propose an authentication scheme to preserve the privacy of user data in edge intelligence-enabled VANETs. The proposed scheme can establish a secure communication channel using fuzzy extractor, elliptic curve cryptography (ECC), and physical unclonable function (PUF) technology. The proposed data upload process can provide privacy of the data using local differential privacy and symmetric key encryption. We validate the security robustness of the proposed scheme using informal analysis, the Real-Or-Random (ROR) model, and the Scyther tool. Moreover, we evaluate the computation and communication efficiency of the proposed and related schemes using Multiprecision Integer and Rational Arithmetic Cryptographic Library (MIRACL) software development kit (SDK). We simulate the practical deployment of the proposed scheme using network simulator 3 (NS-3). Our results show that the proposed scheme has a performance improvement of 10∼48% compared to the state-of-the-art research. Thus, we can demonstrate that the proposed scheme provides comprehensive and secure communication for data management in edge intelligence-enabled VANET environments.

Real-time topology optimization based on multi-scale convolutional attention mechanism

This article proposes a deep learning model based on a multi-scale convolutional attention mechanism network SegNext for improving Topology Optimization (TO-Next), aimed at addressing the computational challenges associated with finite element iterations in traditional density-based methods. TO-Next is trained on diverse topology structures with varying loads, constraints and enriched physical information after initial iterations to enhance its generalization. Following training on three distinct decoder architectures, the optimal encoder–decoder network structure is determined. This study also investigates the algorithm's generalization capability under both single and multiple load constraints. The results indicate the superior performance and efficiency of the proposed method compared with the Solid Isotropic Material with Penalization (SIMP) method, enabling real-time generation of near-optimal topology structures within a short timeframe.

Central Kurdish Text-to-Speech Synthesis with Novel End-to-End Transformer Training

Recent advancements in text-to-speech (TTS) models have aimed to streamline the two-stage process into a single-stage training approach. However, many single-stage models still lag behind in audio quality, particularly when handling Kurdish text and speech. There is a critical need to enhance text-to-speech conversion for the Kurdish language, particularly for the Sorani dialect, which has been relatively neglected and is underrepresented in recent text-to-speech advancements. This study introduces an end-to-end TTS model for efficiently generating high-quality Kurdish audio. The proposed method leverages a variational autoencoder (VAE) that is pre-trained for audio waveform reconstruction and is augmented by adversarial training. This involves aligning the prior distribution established by the pre-trained encoder with the posterior distribution of the text encoder within latent variables. Additionally, a stochastic duration predictor is incorporated to imbue synthesized Kurdish speech with diverse rhythms. By aligning latent distributions and integrating the stochastic duration predictor, the proposed method facilitates the real-time generation of natural Kurdish speech audio, offering flexibility in pitches and rhythms. Empirical evaluation via the mean opinion score (MOS) on a custom dataset confirms the superior performance of our approach (MOS of 3.94) compared with that of a one-stage system and other two-staged systems as assessed through a subjective human evaluation.