Machine Learning based Path Prediction System - Adapting One Model for All Intersections

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：19

、訪客IP：3.15.182.41

姓名

黃楷棋(Kai-Qi Huang) 查詢紙本館藏

畢業系所

資訊工程學系

論文名稱

(Machine Learning based Path Prediction System - Adapting One Model for All Intersections)

相關論文

★ Dynamic Overlay Construction for Mobile Target Detection in Wireless Sensor Networks	★ 車輛導航的簡易繞路策略
★ 使用傳送端電壓改善定位	★ 利用車輛分類建構車載網路上的虛擬骨幹
★ Why Topology-based Broadcast Algorithms Do Not Work Well in Heterogeneous Wireless Networks?	★ 針對移動性目標物的有效率無線感測網路
★ 適用於無線隨意網路中以關節點為基礎的分散式拓樸控制方法	★ A Review of Existing Web Frameworks
★ 將感測網路切割成貪婪區塊的分散式演算法	★ 無線網路上Range-free的距離測量
★ Inferring Floor Plan from Trajectories	★ An Indoor Collaborative Pedestrian Dead Reckoning System
★ Dynamic Content Adjustment In Mobile Ad Hoc Networks	★ 以影像為基礎的定位系統
★ 大範圍無線感測網路下分散式資料壓縮收集演算法	★ 車用WiFi網路中的碰撞分析

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

[檢視]

[下載]

本電子論文使用權限為同意立即開放。
已達開放權限電子全文僅授權使用者為學術研究之目的，進行個人非營利性質之檢索、閱讀、列印。
請遵守中華民國著作權法之相關規定，切勿任意重製、散佈、改作、轉貼、播送，以免觸法。

摘要(中)

在各式的道路環境下，十字路口是其中非常容易引發交通事故的地點。為了降低路口事故的發生率並增進道路安全，本篇論文提出一車載路徑預測系統。當車載行駛接近至十字路口，該系統能及時預測車輛的未來路徑，以即時警示駕駛事故發生的可能性。本篇論文採用簡易且價格低廉的GPS感測器採收車載行駛至十字路口的軌跡資料，並將軌跡資料中的方位角轉換為偏轉角，再整合車速與偏轉角來產生訓練資料。與此同時，我們實作了Borderline-SMOTE來增加少數類別的占比，解決類別不平衡問題。在本篇的演算法中，我們使用隨機森林與自適應增強兩種集成學習演算法來訓練模型，並且實作交叉驗證來降低過適問題的發生機率。最後的實驗結果指出，隨機森林的效能優於自適應增強，而自適應增強的效能優於基礎分類器(決策樹)。

摘要(英)

Intersections have long been known as the place where major traffic accidents most likely to happen. To reduce the number of accidents, this thesis proposes a vehicle path prediction system to predict the future direction when a vehicle is about to cross an intersection. The simple and inexpensive sensor, such as GPS sensor, is used to collect the dataset of vehicle trajectories at intersections. The trend of vehicle movements are derived from the heading in the trajectories, which is then combined with the vehicle speed to generate training data. To deal with the class imbalance problem, Borderline-SMOTE is implemented to increase the proportion of the minority class. In our path prediction algorithm, two ensemble learning algorithms, i.e., Random Forests and AdaBoost, are adopted for model training . The cross validation technique is used to avoid the issue of overfitting. The experiment results indicate that the Random Forest algorithm exhibits the best performance, and the Adaboost algorithm performs better than the base learner (i.e., Decision Tree).

關鍵字(中)

★ 路徑預測
★ 機器學習

關鍵字(英)

論文目次

1 Introduction 1
2 RelatedWork 4
2.1 NonImage-basedmethods........................... 4
2.1.1 GPS................................... 4
2.1.2 IMU................................... 5
2.1.3 DigitalMaps.............................. 6
2.2 Image-basedmethods.............................. 6
3 Preliminary 9
3.1 MachineLearning................................ 9
3.1.1 SupervisedLearning.......................... 9
3.1.2 UnsupervisedLearning......................... 13
3.1.3 ReinforcedLearning.......................... 14
3.2 ImbalancedData................................ 15
4 Design 18
4.1 DataCollection................................. 19
4.2 InternalCalculation............................... 20
4.3 PathPredictionAlgorithm........................... 22
4.3.1 DataExtraction............................. 22
4.3.2 DataPreprocessing........................... 23
4.3.3 ModelGeneration............................ 24
4.3.4 VehiclePathPrediction........................ 27
5 Performance 28
5.1 ExperimentData................................ 28
5.2 EvaluationMeasures.............................. 29
5.2.1 ConfusionMatrix............................ 30
5.2.2 ROCcurve............................... 30
5.2.3 F-measure................................ 32
5.3 PerformanceAnalysis.............................. 32
6 Conclusions 41
Reference 42

參考文獻

[1] Thespecicationofgpssensor. https://www.itri.org.tw/chi/Content/
MsgPic01/Contents.aspx?SiteID=1&MmmID=620624054325020775&MSid=
62063464.
[2] AngelosAmditis,ArisPolychronopoulos,IoannisKaraseitanidis,GeorgeKatsoulis,
and EvangelosBekiaris.Multiplesensorcollisionavoidancesystemforautomotive
applications usinganimmapproachforobstacletracking.In Information Fusion,
2002. ProceedingsoftheFifthInternationalConferenceon, pages812{817.IEEE,
2002.
[3] PeterLBartlettandMikhailTraskin.Adaboostisconsistent. Journal ofMachine
LearningResearch, pages2347{2368,2007.
[4] D. Caveney.Vehicularpathpredictionforcooperativedrivingapplicationsthrough
digital mapanddynamicvehiclemodelfusion.In 2009 IEEE70thVehicularTech-
nologyConferenceFall, 2009.
[5] M EmreCelebi,HassanAKingravi,andPatricioAVela.Acomparativestudyof
ecientinitializationmethodsforthek-meansclusteringalgorithm. Expertsystems
with applications, pages200{210,2013.
[6] YongquanChen,YuandongSun,NingDing,WingKwongChung,HuihuanQian,
GuoqingXu,andYangshengXu.Areal-timevehiclesafetysystem.In System
Integration(SII),2012IEEE/SICEInternationalSymposiumon, pages957{962.
IEEE, 2012.
[7] CorinnaCortesandVladimirVapnik.Support-vectornetworks. Machine learning,
pages 273{297,1995.
[8] NachiketDeo,AkshayRangesh,andMohanMTrivedi.Howwouldsurroundvehicles
move?auniedframeworkformaneuverclassicationandmotionprediction. arXiv
preprintarXiv:1801.06523, 2018.
[9] CharlesElkan.Usingthetriangleinequalitytoacceleratek-means.In Proceedingsof
the 20thInternationalConferenceonMachineLearning(ICML-03), pages147{153,
2003.
[10] TomFawcett.Anintroductiontorocanalysis. Pattern recognitionletters, pages
861{874, 2006.
[11] Hui Han,Wen-YuanWang,andBing-HuanMao.Borderline-smote:anewover-
sampling methodinimbalanceddatasetslearning.In International Conferenceon
IntelligentComputing, pages878{887.Springer,2005.
[12] DavidJHandandRobertJTill.Asimplegeneralisationoftheareaundertheroc
curveformultipleclassclassicationproblems. Machine learning, pages171{186,
2001.
[13] TinKamHo.Randomdecisionforests.In Documentanalysisandrecognition,1995.,
proceedingsofthethirdinternationalconferenceon, pages278{282.IEEE,1995.
[14] JihuaHuangandHan-ShueTan.Vehiclefuturetrajectorypredictionwithadgps/ins-
based positioningsystem.In 2006 AmericanControlConference, pages6pp.{,2006.
[15] AndreasJahn,KlausDavid,andSebastianEngel.5g/ltebasedprotectionofvulner-
able roadusers:Detectionofcrossingacurb.In VehicularTechnologyConference
(VTC Fall),2015IEEE82nd, pages1{5.IEEE,2015.
[16] PrnayJain,ShubhamVarma,HariPrabhatGupta,TanimaDutta,etal.Asu-
pervisedapproachtowardsnetworkcontrolsystemmodelling.In Communication
Systems andNetworks(COMSNETS),20179thInternationalConferenceon, pages
346{353. IEEE,2017.
[17] ChangMookKang,SooJungJeon,Seung-HiLee,andChungChooChung.Para-
metric trajectorypredictionofsurroundingvehicles.In VehicularElectronicsand
Safety (ICVES),2017IEEEInternationalConferenceon, pages26{31.IEEE,2017.
[18] W. Kim,C.M.Kang,Y.S.Son,S.H.Lee,andC.C.Chung.Vehiclepathprediction
using yawaccelerationforadaptivecruisecontrol. IEEE TransactionsonIntelligent
TransportationSystems, pages1{12,2018.
[19] M. Liebner,M.Baumann,F.Klanner,andC.Stiller.Driverintentinferenceaturban
intersectionsusingtheintelligentdrivermodel.In 2012 IEEEIntelligentVehicles
Symposium, pages1162{1167,2012.
[20] MartinLiebner,MichaelBaumann,FelixKlanner,andChristophStiller.Driver
intentinferenceaturbanintersectionsusingtheintelligentdrivermodel.In Intelligent
VehiclesSymposium(IV),2012IEEE, pages1162{1167.IEEE,2012.
[21] WangLinglin,LiuYongxin,andZhaiXiaoke.Designofreinforcelearningcontrol
algorithm andveriedininvertedpendulum.In ControlConference(CCC),2015
34th Chinese, pages3164{3168.IEEE,2015.
[22] Rushi LongadgeandSnehalataDongre.Classimbalanceproblemindatamining
review. arXiv preprintarXiv:1305.1707, 2013.
[23] IvanPisa,GuillemBoquet,JoseLopezVicario,AntoniMorell,andJavierSerrano.
Vaima:Av2vbasedintersectiontracmanagementalgorithm.In WirelessOn-
demand NetworkSystemsandServices(WONS),201814thAnnualConferenceon,
pages 125{128.IEEE,2018.
[24] Aris Polychronopoulos,UllrichScheunert,andFabioTango.Centralizeddatafusion
for obstacleandroadborderstrackinginacollisionwarningsystem.In International
ConferenceonInformationFusion, pages210{216,2004.
[25] DavidMartinPowers.Evaluation:fromprecision,recallandf-measuretoroc,in-
formedness, markednessandcorrelation.2011.
[26] SukmawatiAnggraeniPutrietal.Combiningintegretedsamplingtechniquewith
feature selectionforsoftwaredefectprediction.In CyberandITServiceManagement
(CITSM), 20175thInternationalConferenceon, pages1{6.IEEE,2017.
[27] J. RossQuinlan.Inductionofdecisiontrees. Machine learning, pages81{106,1986.
[28] LawrenceRRabiner.Atutorialonhiddenmarkovmodelsandselectedapplications
in speechrecognition. ProceedingsoftheIEEE, pages257{286,1989.
[29] HikaruShimizu,M-AKobayashi,HiroshiIshikawa,andHarukoFujii.Trac
ow
controlsysteminurbanroadnetworks.In CircuitsandSystems,2004.MWSCAS′04.
The 200447thMidwestSymposiumon, pagesiii{149.IEEE,2004.
[30] MarinaSokolovaandGuyLapalme.Asystematicanalysisofperformancemeasures
for classicationtasks. Information Processing&Management, pages427{437,2009.
[31] YoungSeopSon,WonheeKim,Seung-HiLee,andChungChooChung.Asynchronous
sensor fusionusingmulti-ratekalmanlter.,pages1551{1558,2014.
[32] H. S.TanandJ.Huang.Dgps-basedvehicle-to-vehiclecooperativecollisionwarning:
Engineering feasibilityviewpoints. IEEE TransactionsonIntelligentTransportation
Systems, pages415{428,Dec2006.
[33] Bo Tang,SalmanKhokhar,andRakeshGupta.Turnpredictionatgeneralizedin-
tersections. In IntelligentVehiclesSymposium(IV),2015IEEE, pages1399{1404.
IEEE, 2015.
[34] SebastianThrun,MikeMontemerlo,HendrikDahlkamp,DavidStavens,Andrei
Aron, JamesDiebel,PhilipFong,JohnGale,MorganHalpenny,GabrielHomann,
et al.Stanley:Therobotthatwonthedarpagrandchallenge. Journal ofeld
Robotics, pages661{692,2006.
[35] ShiuHangTsang,EdwardGHoare,PeterSHall,andNigelJClarke.Automotive
radar imageprocessingtopredictvehicletrajectory.In Image Processing,1999.ICIP
99. Proceedings.1999InternationalConferenceon, pages867{870.IEEE,1999.
[36] M. Xiao,Y.Yin,Y.Zhou,andS.Pan.Researchonimprovementofapriorialgorithm
based onmarkedtransactioncompression.In 2017 IEEE2ndAdvancedInformation
Technology,ElectronicandAutomationControlConference(IAEAC), pages1067{
1071, 2017.
[37] J. Yang,H.Bao,N.Ma,andZ.Xuan.Analgorithmofcurvedpathtrackingwith
prediction modelforautonomousvehicle.In 2017 13thInternationalConferenceon
Computational IntelligenceandSecurity(CIS), pages405{408,2017.
[38] Xu Yuan,LucianBusoniu,andRobertBabuska.Reinforcementlearningforelevator
control. IFACProceedingsVolumes, pages2212{2217,2008.
[39] S. Zhao,Y.Chen,andJ.A.Farrell.High-precisionvehiclenavigationinurbanenvi-
ronmentsusinganmem′simuandsingle-frequencygpsreceiver. IEEE Transactions
on IntelligentTransportationSystems, pages2854{2867,2016.

指導教授

孫敏德(Min-Te Sun)

審核日期

2018-7-18

推文