Browsing by Author "Boker, Almuatazbellah M."
Now showing 1 - 6 of 6
Results Per Page
Sort Options
- Cooperative Payload Transportation by UAVs: A Model-Based Deep Reinforcement Learning (MBDRL) ApplicationKhursheed, Shahwar Atiq (Virginia Tech, 2024-08-20)We propose a Model-Based Deep Reinforcement Learning (MBDRL) framework for collaborative paylaod transportation using Unmanned Aerial Vehicles (UAVs) in Search and Rescue (SAR) missions, enabling heavier payload conveyance while maintaining vehicle agility. Our approach extends the single-drone application to a novel multi-drone one, using the Probabilistic Ensembles with Trajectory Sampling (PETS) algorithm to model the unknown stochastic system dynamics and uncertainty. We use the Multi-Agent Reinforcement Learning (MARL) framework via a centralized controller in a leader-follower configuration. The agents utilize the approximated transition function in a Model Predictive Controller (MPC) configured to maximize the reward function for waypoint navigation, while a position-based formation controller ensures stable flights of these physically linked UAVs. We also developed an Unreal Engine (UE) simulation connected to an offboard planner and controller via a Robot Operating System (ROS) framework that is transferable to real robots. This work achieves stable waypoint navigation in a stochastic environment with a sample efficiency following that seen in single UAV work. This work has been funded by the National Science Foundation (NSF) under Award No. 2046770.
- Electromechanical Wave Propagation AnalysisYarahmadi, Somayeh (Virginia Tech, 2024-01-09)When a power system is subjected to a disturbance, the power flow changes, leading to deviations in the synchronous generator rotor angles. The rotor angle deviations propagate as electromechanical waves (EMWs) throughout the power system. These waves became observable since the development of synchrophasor measurement instruments. The speed of EMW propagation is hundreds of miles per second, much less than the electromagnetic wave propagation speed, which is the speed of light. Recently, with the development of renewable energy resources and a growth in using HVDC and FACTS devices, these waves are propagating slower, and their impacts are more considerable and complicated. The protection system needs a control system that can take suitable action based on local measurements to overcome the results of power system faults. Therefore, the dynamic behavior of power systems should be properly observed. The EMW propagation in the literature was studied using assumptions such as constant voltage throughout the entire power system and zero resistances and equal series reactances for the transmission lines. Although these assumptions help simplify the power system study model, the model cannot capture the entire power system's dynamic behaviors, since these assumptions are unrealistic. This research will develop an accurate model for EMW propagation when the system is facing a disturbance using a continuum model. The model includes a novel inertia distribution. It also investigates the impacts of voltage changes in the power system on EMW behaviors and when these impacts are negligible. Furthermore, the impacts of the internal reactances of synchronous generators and the resistances of transmission lines on EMW propagation are explored.
- Highly Robust and Efficient Estimators of Multivariate Location and Covariance with Applications to Array Processing and Financial Portfolio OptimizationFishbone, Justin Adam (Virginia Tech, 2021-12-21)Throughout stochastic data processing fields, mean and covariance matrices are commonly employed for purposes such as standardizing multivariate data through decorrelation. For practical applications, these matrices are usually estimated, and often, the data used for these estimates are non-Gaussian or may be corrupted by outliers or impulsive noise. To address this, robust estimators should be employed. However, in signal processing, where complex-valued data are common, the robust estimation techniques currently employed, such as M-estimators, provide limited robustness in the multivariate case. For this reason, this dissertation extends, to the complex-valued domain, the high-breakdown-point class of multivariate estimators called S-estimators. This dissertation defines S-estimators in the complex-valued context, and it defines their properties for complex-valued data. One major shortcoming of the leading high-breakdown-point multivariate estimators, such as the Rocke S-estimator and the smoothed hard rejection MM-estimator, is that they lack statistical efficiency at non-Gaussian distributions, which are common with real-world applications. This dissertation proposes a new tunable S-estimator, termed the Sq-estimator, for the general class of elliptically symmetric distributions—a class containing many common families such as the multivariate Gaussian, K-, W-, t-, Cauchy, Laplace, hyperbolic, variance gamma, and normal inverse Gaussian distributions. This dissertation demonstrates the diverse applicability and performance benefits of the Sq-estimator through theoretical analysis, empirical simulation, and the processing of real-world data. Through analytical and empirical means, the Sq-estimator is shown to generally provide higher maximum efficiency than the leading maximum-breakdown estimators, and it is also shown to generally be more stable with respect to initial conditions. To illustrate the theoretical benefits of the Sq for complex-valued applications, the efficiencies and influence functions of adaptive minimum variance distortionless response (MVDR) beamformers based on S- and M-estimators are compared. To illustrate the finite-sample performance benefits of the Sq-estimator, empirical simulation results of multiple signal classification (MUSIC) direction-of-arrival estimation are explored. Additionally, the optimal investment of real-world stock data is used to show the practical performance benefits of the Sq-estimator with respect to robustness to extreme events, estimation efficiency, and prediction performance.
- Learning-Based Pareto Optimal Control of Large-Scale Systems with Unknown Slow DynamicsTajik Hesarkuchak, Saeed (Virginia Tech, 2024-06-10)We develop a data-driven approach to Pareto optimal control of large-scale systems, where decision makers know only their local dynamics. Using reinforcement learning, we design a control strategy that optimally balances multiple objectives. The proposed method achieves near-optimal performance and scales well with the total dimension of the system. Experimental results demonstrate the effectiveness of our approach in managing multi-area power systems.
- A Multi-Sensor Passive Occupant LocalizationAmbarkutuk, Murat (Virginia Tech, 2024-11-25)Indoor localization has emerged as a critical technology for enhancing the functionality and efficiency of smart environments. This dissertation focuses on vibro-localization, a novel IOL methodology that determines occupant positions by analyzing floor vibrations caused by footfall patterns. Unlike traditional localization techniques that rely on visual or radio-based sensing, vibro-localization leverages accelerometers fixed to the floor to capture vibro-measurements, offering a cost-effective and privacy-preserving alternative. The primary objective of this research is to address significant limitations in existing vibro-localization approaches, including sensor imperfections, measurement uncertainty, and complex wave dynamics. To this end, we develop comprehensive models that characterize both random and systematic errors introduced by accelerometers, integrating these models into the localization framework to enhance accuracy. Furthermore, we quantify the uncertainty in vibro-measurements and elucidate their contribution to localization errors, providing a robust foundation for error mitigation strategies. A key contribution of this work is the introduction of an information-theoretic Byzantine Sensor Elimination (BSE) algorithm. This algorithm assesses the reliability of vibro-measurement vectors by categorizing sensors into consistent and divergent subsets, thereby minimizing the impact of external uncertainties such as reflections and dispersion. Additionally, we propose multi-sensor vibro-localization techniques that aggregate data from multiple accelerometers, enhancing robustness against individual sensor inaccuracies and environmental variabilities. To accurately model wave propagation, this dissertation advances parametric models that account for dispersion, attenuation, and material inhomogeneities in the floor structure. These models facilitate precise occupant localization even with low-spectral resolution in transfer function estimates. Empirical validation using controlled experimental data demonstrates significant improvements in localization accuracy and precision over baseline methods, highlighting the efficacy of the proposed techniques. The outcomes of this research contribute to the development of economically feasible and ethically sound IOL technologies, broadening their applicability across various domains such as smart homes, healthcare, and energy management. By addressing critical challenges in sensor reliability and wave dynamics, this dissertation paves the way for more accurate, reliable, and scalable indoor localization systems.
- Near-Optimal Control of Atomic Force Microscope For Non-contact Mode ApplicationsSutton, Joshua Lee (Virginia Tech, 2022-06-13)A compact model representing the dynamics between piezoelectric voltage inputs and cantilever probe positioning, including nonlinear surface interaction forces, for atomic force microscopes (AFM) is considered. By considering a relatively large cantilever stiffness, singular perturbation methods reduce complexity in the model and allows for faster responses to Van der Waals interaction forces experienced by the cantilever's tip and measurement sample. In this study, we outline a nonlinear near-optimal feedback control approach for non-contact mode imaging designed to move the cantilever tip laterally about a desired trajectory and maintain the tip vertically about the equilibrium point of the attraction and repulsion forces. We also consider the universal instance when the tip-sample interaction force is unknown, and we construct cascaded high-gain observers to estimate these forces and multiple AFM dynamics for the purpose of output feedback control. Our proposed output feedback controller is used to accomplish the outlined control objective with only the piezotube position available for state feedback.