文档名：区域快速优化下的无人机在线轨迹规划方法
摘要：针对传统基于梯度的规划方法需预先构建欧式符号距离场(euclideansigneddis-tancefield,ESDF)导致障碍物信息冗余度高、规划效率受限问题,提出了一种基于区域快速优化的实时轨迹规划方法.所提方法设计碰撞控制点替换策略用于加快碰撞区域的轨迹优化收敛速度从而降低轨迹规划时间,并定义提取与轨迹规划相关的局部障碍物信息方法,避免构建ESDF过程,从而提高规划效率;之后考虑轨迹安全性、平滑性及动态可行性,建立多目标优化函数,进一步优化轨迹.仿真实验表明,该方法可有效实现无人机在线轨迹规划,且与前沿方法相比,轨迹规划时间平均缩短了36.1%,轨迹优化收敛速度平均提高了33.1%,实现了更高效的规划.
Abstract：Inrecentyears,withtherapidincreaseinthedemandofUAVforrescueanddisasterreliefinunknownandcomplexenvironmentssuchasearthquakeruins,firescene,ruggedmountainsandforests,higherrequirementsareputforwardfortheautonomousnavigationofUAV.InordertoensurethattheUAVcanquicklyrespondtounforeseenriskswhenflyingathighspeedinunknownenvironment,theonlinetrajectoryplanningmoduleinautonomousnavigationisveryimportant.Gradientbasedplanningmethodhastheoutstandingadvantagesofhighsuccessrateandfastplanningspeed,andhasgraduallybecomethemainstreammethodofUAVonlinetrajectoryplanning.However,thetraditionalgradientbasedplanningmethodneedstoconstructEuclideanSignedDistanceField(ESDF)inadvance,whichleadstotheproblemsofhighredundancyofobstacleinformationandlimitedplanningefficiency.Tosolvetheseproblems,thispaperproposesanonlinetrajectoryplanningmethodbasedonregionalfastoptimization.OnlinetrajectoryplanningofUAVisgenerallybasedonstateestimationandvoxelmappingmodule.UpdatedmapsandposeinformationofUAVarefedtotrajectorygenerationmoduletogenerateinitialtrajectory,andthenentertrajectoryoptimizationmoduletogenerateoptimaltrajectory,whichissenttotrajectoryserver,andthecorrespondingflightcontrollercancontrolUAV.Inthispaper,inordertotransformthelocalplanninginunknownenvironmentintothelocalfastoptimizationproblemofinitialtrajectory,uniformB-splineisusedtofurtherparameterizetheinitialtrajectory.AccordingtothecurrentmotionstateandenvironmentalinformationofUAV,amoreefficienttrajectoryoptimizationstrategyisdesignedtoquicklyoptimizetheinitialtrajectorytoahigh-qualitytrajectorythatmeetstherequirementsofsafety,smoothnessanddynamicfeasibility.Trajectoryoptimizationisdividedintotwostages:thefirststageisfasttrajectoryoptimizationincollisionarea.Collisiondetectioniscarriedoutcontinuouslyontheinitialtrajectory.ApairofcontrolpointsQinandQoutareusedtorecordthefirstandlastpositionsofeachcollisionareatrajectory,anda"collisionset"Qcolcomposedofcollisioncontrolpointsisfound.Afterwards,theA*pathsearchalgorithmisusedtosearchfortheoptimalpath,whichistofindasafeguidingpathfromQintoQoutandobtainthesetofpathpointsA.TopusheachcollisioncontrolpointQiinthe"collisionset"Qcolawayfromthecurrentobstacleatthefastestspeedandshortestdistance,acollisioncontrolpointreplacementstrategyisproposed.BysearchingforthecorrespondingpathpointAQforeachcollisioncontrolpointQiinthepathpointsetA,thereplacementoperationisperformed,anditisusedasthenewcontrolpointQinew.Duringthereplacementprocess,onlythepositionofthecollisioncontrolpointsontheinitialtrajectorywasadjustedtominimizetheimpactontheentiretrajectory,allowingformoreflexibleadjustmentofthecollisionareatrajectoryandachievingfasttrajectoryoptimization.Thesecondstageismulti-objectivetrajectoryoptimization,whichpreparesforfurthertrajectoryoptimizationbydefiningandextractinglocalobstacleinformationrelatedtotrajectoryplanningandcalculatingcollisioncost.Then,consideringtrajectorysafety,smoothnessanddynamicfeasibility,multi-objectiveoptimizationfunctionisestablishedtofurtheroptimizetrajectory.Simulationresultsshowthatthismethodhasasignificantimprovementinplanningtimecomparedwithexistingalgorithms:inshort-distanceplanninginsimplescenarios,comparedwiththefrontiermethod,thetotalplanningtime,trajectoryinitializationtimeandtrajectoryoptimizationtimearereducedby36.1%,33.1%and37.7%,respectively;Inthelong-distanceplanningundercomplexscenes,comparedwiththeclassicalgradientbasedplanningmethod,thetrajectoryplanningtimeisshortenedby86.56%onaverage,andtheplanningefficiencyisgreatlyimproved.Comparedwiththecutting-edgegradientbasedplanningmethod,theefficiencyoftrajectoryoptimizationisfurtherimprovedasthecomplexityoftheenvironmentbecomeshigher.Andthemethodproposedinthispapercaneffectivelycarryoutonlineplanningindifferentcomplexenvironments,andhasstrongrobustnessandscalability.
作者：唐嘉宁 和雪梅 陈云浩 彭志祥 周思达 Author：TANGJianing HEXuemei CHENYunhao PENGZhixiang ZHOUSida
作者单位：云南民族大学电气信息工程学院,昆明650000;云南民族大学无人自主系统研究院,昆明650000云南民族大学电气信息工程学院,昆明650000云南民族大学无人自主系统研究院,昆明650000
刊名：重庆理工大学学报
Journal：JournalofChongqingInstituteofTechnology
年，卷(期)：2024, 38(3)
分类号：TP242
关键词：轨迹规划 欧式符号距离场 轨迹优化 梯度信息 多目标优化
Keywords：trajectoryplanning ESDF trajectoryoptimization gradientinformation multi-objectiveoptimization
机标分类号：TP391.41TP242TN929.5
在线出版日期：2024年3月25日
基金项目：国家自然科学基金，国家自然科学基金区域快速优化下的无人机在线轨迹规划方法[
期刊论文] 重庆理工大学学报--2024, 38(3)唐嘉宁 和雪梅 陈云浩 彭志祥 周思达针对传统基于梯度的规划方法需预先构建欧式符号距离场(euclideansigneddis-tancefield,ESDF)导致障碍物信息冗余度高、规划效率受限问题,提出了一种基于区域快速优化的实时轨迹规划方法.所提方法设计碰撞控制点替换策...参考文献和引证文献
参考文献
引证文献
本文读者也读过
相似文献
相关博文
区域快速优化下的无人机在线轨迹规划方法 Online trajectory planning method for UAV under regional fast optimization
区域快速优化下的无人机在线轨迹规划方法.pdf
- 文件大小:
- 1.77 MB
- 下载次数:
- 60
-
高速下载
|
|
