סמינר בית הספר להנדסת חשמל ומחשבים 25/06/2025
סמינר מחלקתי
Join Zoom Meeting: https://us02web.zoom.us/j/88598321479?pwd=jr1Xt0T8wVban98cIhwzRhxJNLkd8a.1
Meeting ID: 885 9832 1479
Passcode: 160706
13:10 Yair Dashevsky . Ms.C Student. Supervisor : Dr. Matan Gal Katziri
Subject: Shape estimation and pattern correction of flexible phased arrays using local curvature measurements
Abstract: This work presents a general method to estimate the unknown shapes of mechanically flexible phased arrays using strain gauge-based curvature measurements. Such information is crucial to resolve the positions and orientations of the radiators and perform effective beamforming. The procedure is feedback-free, insensitive to electromagnetic interference, and provides a low-frequency modality of shape information. While not limited to a specific size or frequency of operation, this method is demonstrated using a 20-cm, 1 × 8 phased arrayoperating at 6 GHz, achieving shape reconstruction errors of ≈0.052λ for singly, doubly, and triply bent arrays with curvature radii of less than 3.5 cm. Moreover, we illustrate how the shape information can be utilized to correct the beam pattern of a flexible array and enable beamforming and beam-steering even under severe deformations
Bio: Yair Dashevsky received the B.S degree in Electrical engineering from Ben-Gurion University, Beer-Sheva, Israel, in 2023
He is currently pursuing the M.Sc. degree in electrical engineering in Ben-Gurion University
His research interests include analog integrated circuits, RF circuits, phased arrays and RFIC
13:35 Linoy Ketashvily. Msc. Student Supervisor: Dr. Nir Shlezinger and Prof. Ilana Nisky
Subject: AI-aided Kalman Tracking for Cable-Driven Robots for Surgery
Abstract: Cable-driven robots are widely used robotic systems that utilize cables and winches for precise manipulation and movement. These robots function by using multiple cables attached to a central platform or end-effector, with each cable being controlled by motor and encoders located at the base of the robot and not on the governing joint. This design results in a more lightweight system which is essential for certain applications e.g. minimally invasive surgeries. The actuation of the joints is transmitted through non-linear elements, which cause errors in localization due to the non- linear property of cable tension. In terms of modelling, this induces an unknown time-varying bias term in the observation model governing the system dynamics. In this work, we propose an algorithm which suggests a new approach for overcoming biased dynamic systems when the bias is unknown and accurately track while relying solely on existing biased observations. This work presents a novel data-driven algorithm that addresses these biases in partially known dynamic systems, focusing on cable-driven robots. The proposed approach integrates classic Kalman filtering with deep learning in a supervised manner to enhance tracking and overcome the limitations of existing state-space modeling. Our algorithm learns from data to explicitly track the robot’s end-effector, while implicitly tracking its cable-induced bias
Bio: Linoy Ketashvili is an M.Sc. researcher in the Department of Electrical and Computer Engineering at Ben-Gurion University of the Negev supervised by Dr. Nir Shlezinger and Prof. Ilana Nisky. She holds a B.Sc. in Biomedical Engineering from Tel Aviv University. Her research focuses on AI-aided Kalman tracking for cable-driven robots, which she presented at ICASSP, a leading international conference in signal processing
14:00 Ohad Akler. Ms.C Student. Supervisor: Prof. Alon Kuperman
Subject: Advanced modeling and control of high-power resonant inverters
Abstract: Applications such as hydrogen fusion, magnetic levitation, and induction heating systems employ resonant inverters as means of electrical energy delivery. Such systems are highly non-linear and time-varying, imposing complicated operational challenges. On the other hand, precise control is required to keep these systems within the desired region of operation. Consequently, the proposed research will explore advanced modeling and control techniques as well as power electronics design ensuring optimal performance of resonant inverters employed in hydrogen fusion systems
Bio: Ohad is an M.Sc. student at the Applied Energy Laboratory in the Department of Electrical and Computer Engineering, working under the supervision of Prof. Alon Kuperman
During the research, Ohad has published multiple papers, including an accepted paper to a leading Q1 journal, with another journal paper currently under review
14:25 Ori Kenig. Ms.C Student . Supervisor : prof. Kobi Todros
Subject: Robust Gaussian Mixture Modeling: A K-divergence based approach
Abstract: This thesis addresses the problem of robust Gaussian mixture modeling in the presence of outliers
We commence by introducing a general expectation-maximization (EM)-like scheme, called K-BM, for iterative numerical computation of the minimum K-divergence estimator (MKDE). This estimator leverages Parzen's non-parametric Kernel density estimate to down-weight low density regions associated with outlying measurements. Akin to the conventional EM, the K-BM involves successive Maximizations of lower Bounds on the objective function of the MKDE. However, differently from EM, these bounds are not exclusively reliant on conditional expectations. The K-BM algorithm is applied to robust parameter estimation of a finite-order multivariate Gaussian mixture model (GMM)
We proceed by introducing a new robust variant of the Bayesian information criterion (BIC) that penalizes the objective function of the MKDE. The proposed criterion, called K-BIC, is conveniently applied for robust GMM order selection. In this work, we also establish a data-driven procedure for selection of the kernel's bandwidth parameter. This procedure operates by minimizing an empirical asymptotic approximation of the mean-integrated-squared-error (MISE) between the underlying density and the estimated GMM density
Lastly, the K-BM, the K-BIC, and the MISE based selection of the kernel's bandwidth are combined into a unified framework for joint order selection and parameter estimation of a GMM
The advantages of the K-divergence based framework over other robust approaches are illustrated in simulation studies involving synthetic and real data
Bio: Ori Kenig has an undergraduate degree in Mathematics with Computer Science at the Technion and a Masters degree in Computer Engineering at the American air force institute of technology (AFIT). He is currently working towards completing an Electrical Engineering degree in Ben Gurion University of the Negev (BGU)
