Online parameter identification framework for a multirotor UAV: Application to an arm stretchable morphing multirotor

This paper presents a new framework to identify the dynamic characteristics of a multirotor unmanned aerial vehicle (MUAV). In particular, the identification framework is applied for an arm stretchable morphing MUAV. This MUAV is capable of adjusting the arm length during flight to maximize performance as well as stability. However, it is challenging to develop a flight controller without understanding several parameters of the MUAV, even though it has similar dynamic characteristics compared to conventional MUAVs. The framework consists of three sequentially operated steps: constrained roll dynamics on the ground, yaw dynamics, and roll/pitch dynamics in flight. A system is excited safely by a designed command for each step. Then, the Extended Kalman Filter (EKF) is utilized to estimate the unknown parameters that can best match the motion data with the dynamic model. The identification framework offers an effective means to design the controller by providing its dynamic characteristics. It is validated by numerical simulations and experiments with the arm stretchable MUAV. Furthermore, the identified parameters are utilized to maximize control performance using particle swarm optimization (PSO) in experiments. The flight performance along with the identified parameters has been experimentally demonstrated.