简介：ItisofgreatsignificancetohaveaclearunderstandingoftheinternationalenvironmentthatChinafaces.Butthisisanuneasyjob.SinceSeptember11,theworldsituationhasbecomesocomplicatedthatitishardtograspitsessenceevenwithpainstakingefforts.TojudgesuchanalreadycomplicatedinternationalenvironmentfromChina’sperspectiveaddsmoredifficulties,fordifferentpeopleusuallyhavedifferentviewsduetothedifferentanglesfromwhichtheyseethesituationanddifferentmethodstheyadopt.This,understandably,leadstohotdebateswithintheChineseacademiccircles.Toprovideastageforexchangeofviews,attheinvitationoftheCIRExpertForum,somewell-knownscholarsandexpertsoninternationalrelationsfromBeijinggatheredtoholdasymposiumonNovember5,2002,withthethemeof"HowtoassesstheinternationalenvironmentthatChinafaces".Thespeechesatthesymposiumarepublishedintwoinstallments.Followingarethesecondhalf.

简介：Theperformanceofdeeplearning(DL)networkshasbeenincreasedbyelaboratingthenetworkstructures.However,theDLnetowrkshavemanyparameters,whichhavealotofinfluenceontheperformanceofthenetwork.Weproposeageneticalgorithm(GA)baseddeepbeliefneuralnetwork(DBNN)methodforrobotobjectrecognitionandgraspingpurpose.ThismethodoptimizestheparametersoftheDBNNmethod,suchasthenumberofhiddenunits,thenumberofepochs,andthelearningrates,whichwouldreducetheerrorrateandthenetworktrainingtimeofobjectrecognition.Afterrecognizingobjects,therobotperformsthepick-andplaceoperations.Webuildadatabaseofsixobjectsforexperimentalpurpose.Experimentalresultsdemonstratethatourmethodoutperformsontheoptimizedrobotobjectrecognitionandgraspingtasks.

简介：[摘要] 本论文中提出的机器视觉四轴机器人抓取系统应用于自动化生产车间中，首先采用机器二维视觉对零件拍照，通过视觉算法分析零件的角度、中心点、位置、颜色等关键信息，视觉控制器将分析结果以字符串的形式发送给四轴机器人控制器，机器人控制器获得产品零件的几何中心、零件外形、产品颜色和零件旋转角度信息，然后机器系统规划出零件的最优抓取路径和零件位置，最后控制机器人末端抓取零件，放置待料工位，实现零件的自动化上料、抓取、放置等操作。