Battleship but 2 steps ahead
Trajectory Research: Analysis, Classification, and Knowledge-building
In numerous applications (GPS navigation, object tracking using radar systems, vehicle tracking in videos), the analysis, estimation, and classification of object trajectories from noisy observations play a key role. Studying trajectories enables the detection of abnormal ship trajectories, as well as satellite trajectories, the understanding of certain convoy strategies in operational contexts, and the formulation of hypotheses on the short-term movements of low-maneuverability objects such as container ships, etc.
Project Team Structure
The project team is structured around complementary roles covering signal processing, artificial intelligence, and communication. Under the coordination of project leader Milo, the team is divided into several expert groups: signal processing (Arnaud, Angel, Idriss) works on real-time signal characterization methods; the LSTM group (Benjamin, Yassine, Rayane) focuses on developing and optimizing recurrent neural network architectures for trajectory classification; the CNN team (Milo, Samy) handles convolutional approaches applied after signal preprocessing; finally, the communications section (Raphael) is in charge of developing a website to support project dissemination, interactive demonstrations, and promotion materials.
