Stochastic Model Predictive Control for the Set Point Tracking of Unmanned Surface Vehicles

Yuan Tan,Kok Lay Teo,Song Wang,Guangbin Cai,Bin Li

doi:10.1109/access.2019.2962061

Abstract

An unmanned surface vehicles (USV) set point tracking problem is investigated in this paper. The stochastic model predictive control (SMPC) scheme is utilized to design the controller in order to reject the environment disturbances and meet the physical constraints. The design problem is formulated as a chance-constrained stochastic optimization problem, which is non-convex. Thus, the problem is computationally prohibitive. For this, the convex conditional value at risk (CVaR) approximation is introduced to convert the chance constraints into deterministic convex constraints. The converted constraints are then further transformed into the second order cone (SOC) constraints. Therefore, the proposed method is computationally tractable and hence can be implemented online. A numerical example is provided to illustrate the effectiveness of the proposed method.

Highlights

The unmanned surface vehicle (USV) is a powerful tool for commercial, scientific and military applications since it is cheaper, safer and more flexible than the commercial ship
The rest of this paper is structured as follows: The state-space model of the USV is established in Section 2; An online stochastic model predictive control (SMPC) algorithm is developed in Section 3; Simulation results are reported in Section 4 to illustrate the effectiveness of the proposed algorithm, and Section 5 concludes this paper
Inspired by [30], these conditional value at risk (CVaR) approximations are further converted into tractable second order cone (SOC) constraints

Summary

INTRODUCTION

The unmanned surface vehicle (USV) is a powerful tool for commercial, scientific and military applications since it is cheaper, safer and more flexible than the commercial ship. The applications of USVs are facing great challenge in the control system design. This is because the USVs are usually working in the environment with external disturbances. The MPC scheme is utilized to solve an USV set point tracking problem. The set point tacking problem is formulated as a chance-constrained stochastic model predictive control (SMPC) problem. This is mainly because the chance constraints are non-convex and are computationally intractable. In SMPC, this scheme is called scenario generation method [31] The problem of this method is that the computation burden is high, which cannot be implemented online. The rest of this paper is structured as follows: The state-space model of the USV is established in Section 2; An online SMPC algorithm is developed in Section 3; Simulation results are reported in Section 4 to illustrate the effectiveness of the proposed algorithm, and Section 5 concludes this paper

THE SYSTEM MODEL OF USV

PROBLEM STATEMENT

SIMULATION RESULTS

CONCLUSION