Gaussian Process Model Predictive Control of unmanned quadrotors

Abstract

Two issues of quadrotor control without deterministic dynamical equations are addressed in this paper by using Gaussian Process (GP) based Model Predictive Control (MPC) algorithm. Firstly, the first issue of modelling unknown dynamical motions is solved by using GP models based on sampled data. In this way, the model uncertainty can be numerically evaluated during modelling and prediction process. This is not easy when using other data-driven methods, such as Artificial Neural Networks (ANN) and Fuzzy Models (FMs). Then a MPC scheme based on obtained GP models is proposed to address the second issue of designing appropriate quadrotor controllers. The proposed algorithm directly takes model uncertainty into account when planning MPC policies, and can be computationally efficiently implemented through using analytical gradients in the optimization process. The performance of quadrotor control using proposed approach is demonstrated by simulations on a trajectory tracking problem.

Cautious Model Predictive Control Using Gaussian Process Regression

“…In [10] a piecewise linear approximate explicit solution for the MPC problem of a combustion plant was presented. Application of a one-step MPC with a GP model to a mechatronic system was demonstrated in [11] and the use for fault-tolerant MPC was presented in [12]. A constrained tracking MPC for robotic applications, which uses a GP to improve the dynamics model from measurement data, was shown in [13].…”

A constrained instantaneous learning approach for aerial package delivery robots: onboard implementation and experimental results

“…The disturbances are modeled as GPs, which actually represent the discrepancies between the a priori model and the measured system behavior. In addition, (Cao et al, 2016) implemented the ILC technique for the control of an unmanned quadrotor. In their application, the whole quadrotor dynamics is modeled using GP and is learned from the beginning, purely utilizing observations.…”

Networked model predictive control using a wavelet neural network

“…Cao et al 30 used a Gaussian process model of the unknown dynamics of a quadrotor with model predictive control. Their methodology handles the model uncertainty and is computationally efficient.…”