This research explores cutting-edge deep learning techniques, focusing on the development and optimization of advanced neural network architectures to address complex, real-world problems. Key areas include deep reinforcement learning (DRL), convolutional neural networks (CNNs), transformers, and emerging models such as Mamba and Physics-Informed Neural Networks (PINNs). These methodologies are used to create efficient, adaptive controllers for specialized tasks and to autonomously extract meaningful patterns from diverse data sources, including radar, images, videos, and time-series signals.

Applications: space mission, planetary exploration, autonomous flight control, object detection and tracking, path planning, model-free control, electronic warfare modeling, time-series analysis, multi-agent systems.

This research focuses on advancing optimization techniques, emphasizing both theoretical foundations and practical, real-time implementations. The core objective is to develop and refine methods for efficiently solving complex optimization problems, particularly in the context of model predictive control (MPC). Current efforts center on integrating machine learning-enhanced optimization, combining deep learning with traditional optimization to improve speed and adaptability, and accelerating MPC through online optimization by leveraging adaptive algorithms and computational efficiencies to meet real-time demands.

Applications: space orbit design, envelope protection control, skip trajectory optimization, missile guidance, robotics and autonomous systems, wind farm control.

This research investigates advanced nonlinear filtering techniques, including extended Kalman filters (EKF), unscented Kalman filters (UKF), particle filters (PF), interacting multiple model (IMM) filters, and probabilistic data association filters (PDAF). The focus is on developing learning-augmented filters by integrating deep learning with traditional filtering techniques to enhance adaptability and performance in complex environments. Additionally, the research explores distributed filtering for multi-agent systems, quantum-inspired filtering for next-generation sensing applications, and methods to improve computational efficiency for edge devices.

Applications: multi-sensor fusion, target tracking and data association, navigation and positioning system, fault detection and isolation, model identification.