An innovative approach of gaining insight into motor skills involved in human body flight is proposed. The key idea is the creation of a model autonomous system capable of virtually performing skydiving maneuvers. A dynamic skydiver model and simulator is developed, comprising biomechanical, aerodynamic, and kinematic models, dynamic equations of motion, and a virtual reality environment. Limb relative orientations, and resulting inertial body angular position and velocity are measured in skydiving experiments in a vertical wind tunnel and in free fall. These experimental data are compared with corresponding simulation data to tune and verify the model for basic skydiving maneuvers. The model is further extended to reconstruct advanced aerial maneuvers, such as transitions between stable equilibria. The experimental data are used to estimate skydiver's conscious inputs as a function of time, via an Unscented Kalman Filter modified for this purpose.