Koopman Theory in Deep Learning for Linearizing Drone Dynamics

In future, Drones are expected to be used in many scenarios which demands high speed and precise maneuvering and control. Although drones getting upgraded from hardware side, for them to be faster, the optimization algorithm itself is slow because of non linear nature of drone state space.

In this project, I worked on combining Koopman Theory with Deep Learning techniques to get a drone state space equation that is linear at all points and can directly be used in Linear Constrained MPC for faster path planning and other use cases.

The code and final report will be updated soon