Center of Gravity Estimation for Rotary-Wing Aircraft Using Neural Networks

This work shows that a recursive neural network can be used to provide a categorical prediction of the Center of Gravity (CG) for a unmanned aerial vehicle (UAV) helicopter. A SAB Goblin Raw 420 UAV, with a Pixhawk 6C mini controller was modified to allow for lateral and longitudinal shifts in CG. Flight data from the controller was recorded and used to train a Long Short-Term Memory neural network to estimate and classify the CG location for a given flight. Test results show that the neural network is able to effectively identify the correct CG location.