基于監(jiān)督式DDPG算法的小型ROV運動控制方法

打開文本圖片集
本文引用格式:,,,等.基于監(jiān)督式 DDPG 算法的小型ROV 運動控制方法[J].自動化與信息工程,2025,46(3):23-29.HUANG Zhaojun, ZHANG Yanjia, ZUO Xiaowen, et al. Motion control method for small ROV based on super-vised DDPG algorithm[J]. Automation & Information Engineering,2025,46(3):23-29.
關(guān)鍵詞:監(jiān)督式DDPG;小型ROV;運動控制;專家經(jīng)驗;強化學(xué)習(xí)中圖分類號:TP242.3 文獻標(biāo)志碼:A 文章編號:1674-2605(2025)03-0004-07DOI: 10.12475/aie.20250304 開放獲取
Motion Control Method for Small ROV Based on Supervised DDPG Algorithm
HUANG Zhaojun ZHANG Yanjia ZUO Xiaowen CHEN Zexun (Zhuhai City Polytechnic, Zhuhai 519090, China)
Abstract:To adressthe issues of prolonged learning time and diffculty inconvergence when using the Deep Deterministic Policy Gradient (DDPG)algorithm formotioncontrolofremotelyoperatedtetheredunderwatervehicles (ROVs),this paper proposesasupervisedDDPG-basedmotioncontrolmethodforsmallROVs.During theinitialleamingphaseoftheDDPGalgorith, asupervisedleamingapproachisintroduced toaccelerateneural networkconvergenceandreduceleaingtimebyleveragingexpert experience.Simulationresults demonstrate that te supervised DDPGalgorithmachieves superiorcontrolperformancecomparedto the standard DDPG algorithm.
Keywords: supervised DDPG; small ROV; motion control; expert experience; reinforcement learning
0 引言
無人遙控有纜水下機器人(remotelyoperatedvehicle,ROV)因在開發(fā)難度、研制周期、資金投入和產(chǎn)業(yè)化等方面具有優(yōu)勢,成為水下機器人領(lǐng)域的研究重點,并己廣泛應(yīng)用于海水養(yǎng)殖、海洋勘探、水下搜救和石油開發(fā)等領(lǐng)域。(剩余7710字)