System identification of a dynamical system using a physics-informed neural network (PINN)
In this project, we consider the problem of learning unknown dynamics models of a lunar lander using neural network techniques and partial information of the system dynamics. The motivation is to combine the universal approximation property of neural networks and the physical laws that are applied on the desired system in order to accurately identify the dynamical model from limited or small amount of data. We use randomly generated data from a simple lunar lander simulator and develop different learning algorithms to approximated the dynamics model of the simulator.