Real-time estimation of the mass and inertia tensor of quadrotors for controller mapping

Abstract

The expanding need for UAVs to carry payloads with grasping abilities requires developing and improving the motion control systems of these aerial robots. Providing UAVs, such as quadrotors, with grasping abilities requires their ability to transport payloads in optimal time, minimal energy consumption, in addition to risk lessening in cases of dangerous missions, and many other advantages. This thesis aims at devising an online estimation scheme of the varying quadrotor’s parameters as it picks up and manipulates a payload. In particular, the mass and the moment of inertia tensor of the quadrotor are estimated in real-time, while subjected to an additional payload, by analyzing the acquired input-output data in a direct closed-loop fashion and using its rigid body dynamic model. A modified version of the recursive least squares (RLS) method is leveraged for the proposed real-time estimation in this work. While most existing methods assume symmetry of the inertia tensor matrix and disregard its off-diagonal elements, which affects the control system's performance due to the disparity between the predicted model and the physical plant, in this work all of the inertia tensor parameters along with the quadrotor's changing mass are estimated and passed to the control system to achieve enhanced tracking performance. Covariance resetting is integrated into the estimation algorithm to increase its convergence rate and accuracy. Physical constraints are also used to attain consistent and rational estimates of the inertia tensor matrix. The proposed identification scheme is validated in numerical simulations and experimentation on the Quanser 3 DOF Hover and on a real-life quadrotor, the Quanser QBall-2. The obtained results demonstrate the accuracy and convergence rate of the designed estimator, paving the way in front of its integration into an adaptive control system. This system identification scheme is a main building block for future work that involves a controller mapping scheme, which calculates new control