Real-time Estimation of the Inertia Tensor Elements of a Quadcopter Hover Platform

The ability of unmanned aerial vehicles (UAVs) to carry variable payloads, and to support various gripping attachments for transportation and delivery services, is under increasing demand. This work aims at designing an online estimator that determines the change in the dynamic system parameters of a three degrees-of-freedom (3 DOF) quadcopter hover platform, as a payload is varied. In particular, the moment of inertia tensor of the 3 DOF quadcopter hover platform is estimated in real-time, while subjected to an additional payload, by analyzing the acquired input-output data and using its rigid body dynamic model. A recursive least squares (RLS) estimation method is employed, and covariance resetting is integrated to improve the estimator's convergence rate. Simulation and experimental results conducted on the Quanser 3 DOF Hover platform demonstrate the accuracy of the designed estimator, given its ability to accurately and promptly estimate all nine elements of the 3x3 inertia tensor matrix, even in the presence of sudden payload variations. The developed methodology can be extended to design an adaptive scheme that controls the motion of quadcopter UAVs, and utilized in reliable machine health monitoring algorithms.