Autopilot Design for Unmanned Surface Vehicle based on CNN and ACO

IAENG Internaitonal journal of computer science

Wen Ou¹  Tiantian Yang²  Dongming Zhao²  Hao Zhou² 

[1] Huazhong University ofScience and Technology;Wuhan University of Technology

关键词: USV;    autopilot;    predictive control;    Convolution Neural Network (CNN);    Ant Colony Optimization (ACO);    rolling optimization;   

DOI  :  10.15837/ijccc.2018.3.3236

学科分类：计算机科学（综合）

来源: International Association of Engineers

【 摘 要 】

There is a growing concern to design intelligent controllers for autopiloting unmanned surface vehicles as solution for many naval and civilian requirements. Traditional autopilot’s performance declines due to the uncertainties in hydrodynamics as a result of harsh sailing conditions and sea states. This paper reports the design of a novel nonlinear model predictive controller (NMPC) based on convolutional neural network (CNN) and ant colony optimizer (ACO) which is superior to a linear proportional integral-derivative counterpart. This combination helps the control system to deal with model uncertainties with robustness. The results of simulation and experiment demonstrate the proposed method is more efficient and more capable to guide the vehicle through LOS waypoints particularly in the presence of large disturbances.

【 授权许可】

CC BY