The Differential Dynamic Programming (DDP) is a nowadays quite popular optimal control algorithm, which was introduced by Jacobson all the way back in the 1970s. Recently, it has made a resurgence following work by Tassa and Todorov.
I am a third-year PhD student between the Gepetto team at LAAS-CNRS, and the WILLOW team of Inria Paris, under the supervision of Nicolas Mansard and Justin Carpentier.
Previously, I obtained an engineering degree from École polytechnique, in the Paris area, and a master's degree in applied mathematics and CS (MVA) from École normale supérieure Paris-Saclay.
My work focuses on optimization and control for robotics.
Optimal Transport: Wasserstein distance and Sinkhorn
The goal of optimal transport problems, is to find optimal mappings between probability meaures: these mappings are also called transport plans, and can take the form of functional transforms (in Monge's original problem) or joint probability distributions (in the Kantorovitch relaxation).
Introduction to optimal control: LQR
Most control loops out there used in real-world systems are simple feedback loops proportional to the error, its derivative or integral (this is called PID control). However, this kind of control can exhibit undesirable behavior such as oscillations or failing to converge to a given setpoint quickly if at all. Some more complex systems such robots, satellites or cars can come with precise performance requirements, and more carefully constructed control actions need to be supplied with guarantees about their optimality.