AI - Trajectory Modeling

This section presents a collection of research and applied projects focused on trajectory modeling and optimization across multiple domains🤖🛫🌊🚢, including aviation, robotics, and maritime navigation. The work combines advanced Artificial Intelligence techniques such as imitation learning, reinforcement learning, and heuristic optimization algorithms to model, predict, and optimize movement patterns under complex environmental conditions.

Core methodologies include Variational Autoencoders (VAE) for latent feature extraction, Directed-Info GAIL for learning multi-modal trajectory policies from unsegmented data, and graph-based approaches such as the A* algorithm for route optimization. These techniques are applied to real-world and simulated datasets, incorporating contextual factors such as meteorological conditions, system dynamics, and operational constraints.

The projects demonstrate the ability to generate realistic trajectories, identify behavioral modes, and optimize routes from initial states to target goals. Applications range from aircraft flight path prediction and robotic motion control to ship weather routing using CMEMS data through tools such as SIMROUTE.

Overall, this work highlights the integration of data-driven learning and physics-informed modeling to address complex trajectory problems, offering scalable solutions for prediction, decision-making, and optimization in dynamic environments.

Click on each image or video below to view the project’s description and abstract.