基于模型预测控制的无人艇VO避障  

Obstacle avoidance method based on model predictive control combined with velocity obstacle for unmanned surface vessels

作　　者：钟雨轩[1,2] 陈晓虎 马向峰 ZHONG Yu-xuan;CHEN Xiao-hu;MA Xiang-feng(School of computer engineering and science,ShangHai University,ShangHai 200444,China;CSSC systems engineering research institute,BeiJing 100094,China)

机构地区：[1]上海大学机电工程与自动化学院,上海200444 [2]中国船舶工业系统工程研究院,北京100094

出　　处：《舰船科学技术》2019年第23期147-154,共8页Ship Science and Technology

基　　金：科技部重点研发计划“自组网海洋环境多参数测量仪”(2018YFF0103400)

摘　　要：在高速行驶中,无人艇速度快、反应时间短,为了达到安全避障的要求,避障算法必须实时性好、稳定性高。同时,在高速场景中,无人艇的静浮力较小,流体升力较大,风浪流对无人艇的影响较低速场景更加显著,稳定的航迹跟踪与避障控制面临挑战。为了达到良好的避障效果,无人艇必须要满足控制量的约束,同时要削减外界扰动带来的不利影响。无人艇控制系统是一个多输入多输出系统,而模型预测控制则是处理多变量约束优化问题最有效的方法之一。本文提出基于模型预测控制的无人艇VO避障算法,通过仿真实验实现了该算法的避障功能,并将该算法与传统的VO算法避障效果作对比,验证了该算法的有效性,同时,通过分析模型不确定性对该避障算法性能的影响,证明了该算法的鲁棒性。Unmanned surface vessels(USV)are important equipment to ensure the national maritime security.It is necessary for USV to avoid obstacles in congest environment.However,due to the high speed and the short reaction time in high speed status of USV,the obstacle avoidance algorithm must perform real time and stability characteristic in order to avoid obstacles safely.Meanwhile,the static buoyancy of USV decreases and the fluid lift increases during high speed status,the effect of wind and waves becomes more significant compared with low-speed status,which brings challenges in path following and obstacle avoidance.In order to achieve a better effect in obstacle avoidance,USV must satisfy several constraints and reduce the adverse effects of external disturbances.The control system of USV is a multi-input multi-output system,while model predictive control(MPC)is one of the most effective method to deal with multivariable constrained optimization problems.This paper propose an obstacle avoidance method for USV based on model predictive control combined with velocity obstacle(MPC-VO)method,and build a simulation platform to test the validity of this method.The performance of MPC-VO based method and the tradition method are compared in simulation platform,results show that MPC-VO based method is efficient and feasible.Meanwhile,the robustness of the algorithm is proved by analyzing the influence of model uncertainty on performance of MPC-VO based method.

关 键 词：水面无人艇 避障 模型预测控制 VO 

分 类 号：U664.82[交通运输工程—船舶及航道工程]