Gaussian Process Model Predictive Control of An Unmanned Quadrotor

TL;DR

This paper presents a Gaussian Process-based Model Predictive Control approach for quadrotors that efficiently handles input constraints and reduces computational time compared to traditional NMPC, while maintaining similar tracking performance.

Contribution

The paper introduces a novel GP-based MPC framework for quadrotors that relaxes nonlinear optimization problems to convex ones, enabling faster computation with constraint handling.

Findings

Comparable trajectory tracking to NMPC

20% reduction in computational time

Effective constraint incorporation

Abstract

The Model Predictive Control (MPC) trajectory tracking problem of an unmanned quadrotor with input and output constraints is addressed. In this article, the dynamic models of the quadrotor are obtained purely from operational data in the form of probabilistic Gaussian Process (GP) models. This is different from conventional models obtained through Newtonian analysis. A hierarchical control scheme is used to handle the trajectory tracking problem with the translational subsystem in the outer loop and the rotational subsystem in the inner loop. Constrained GP based MPC are formulated separately for both subsystems. The resulting MPC problems are typically nonlinear and non-convex. We derived 15 a GP based local dynamical model that allows these optimization problems to be relaxed to convex ones which can be efficiently solved with a simple active-set algorithm. The performance of the proposed approach is compared with an existing unconstrained Nonlinear Model Predictive Control (NMPC). Simulation results show that the two approaches exhibit similar trajectory tracking performance. However, our approach has the advantage of incorporating constraints on the control inputs. In addition, our approach only requires 20% of the computational time for NMPC.