氏名　Chu Cong Minh
学位の種類　博士（工学）
学位記番号　博甲第422号
学位授与の日付　平成19年3月26日
学位論文題目　Analysis of Motorcycle Behavior at Midblocks and Signalized Intersections　（単路部と信号交差点における自動二輪車の挙動分析）
論文審査委員
　主査　助教授　佐野可寸志
　副査　教授　松本　昌二
　副査　教授　中出　文平
　副査　助教授　樋口　秀
　副査　京都大学助教授　吉井　稔雄

• 論文目次
• 論文要旨
• 論文審査の結果の要旨

［平成18（2006）年度博士論文題名一覧］ ［博士論文題名一覧］に戻る．

Title Paper　p.i
Abstract　p.ii
Acknowledgement　p.iv
Table of Contents　p.v
List of Figures　p.viii
List of Tables　p.xi

1.Introduction
　1.1　Background　p.1
　1.2　Thesis Objectives　p.3
　1.3　Thesis Contributions　p.5
　1.4　Organization　p.7

2.Literature Review
　2.1　Traffic Operation of Two-wheeled Vehicles　p.8
　2.1.1　Traffic Characteristics of Two-wheeled Vehicles　p.8
　2.1.2　Lane of Two-Wheeled Vehicles　p.11
　2.1.3　Traffic Simulation Considering Two-wheeled Vehicles　p.14
　2.2　Acceleration Models　p.17
　2.2.1　Car Following Models　p.18
　2.2.2　General Acceleration Models　p.19
　2.2.3　Reaction Time Distribution　p.21
　2.2.4　Longitudinal Threshold Distance for Free-flow and Following Regimes　p.24
　2.3　Lane Changing Models　p.28
　2.3.1　Longitudinal Threshold Distance for Lane Changing Model　p.28
　2.3.2　General Models　p.28
　2.3.3　Gap Acceptance Models　p.31
　2.4　Summary　p.33

3.Methodology
　3.1　Dynamic Motorcycle's Lane　p.36
　3.2　Longitudinal Threshold Distance of a Motorcycle　p.37
　3.2.1　Longitudinal Threshold Distance 1　p.38
　3.2.2　Longitudinal Threshold Distance 2　p.38
　3.3　Deceleration Models　p.39
　3.3.1　Free-deceleration Model　p.40
　3.3.2　Following-deceleration Model　p.41
　3.3.3　Likelihood Function Formulation　p.42
　3.4　Acceleration Models　p.43
　3.4.1　Free-acceleration Model　p.44
　3.4.2　Following-acceleration Model　p.45
　3.4.3　Likelihood Function Formulation　p.46
　3.5　Reaction Time Distribution　p.47
　3.6　Maneuver Model　p.47
　3.6.1　Lane Selection Model　p.49
　3.6.2　Gap Acceptance Model　p.50
　3.6.3　Likelihood Function of Lane Selection Model and Gap Acceptance Model　p.51
　3.6.4　Adjacent-lane-acceleration Model　p.52
　3.6.4.1　Free-acceleration Model　p.53
　3.6.4.2　Following Model　p.55
　3.6.4.3　Adjacent-gap-acceleration Model　p.55

4.Data Collection and Analysis
　4.1　Data Requirements　p.57
　4.2　Measuring Devices and Data Collection Methodology　p.58
　4.3　Data for Midblock Analysis　p.59
　4.4　Data for Signalized Intersection Analysis　p.61
　4.5　Data Extraction　p.63
　4.6　Data Analysis by Using SEV Software　p.64
　4.6.1　Coordinate Transformation Technique　p.65

5.The Midblock Analyses
　5.1　Estimation of Motorcycle Units （MCUs）　p.68
　5.1.1　Effective Space of a Vehicle　p.68
　5.1.2　Formula of Motorcycle Units （MCUs）　p.72
　5.1.3　Validation of Motorcycle Units （MCUs）　p.72
　5.2　Speed Distribution　p.76
　5.3　Headway Distribution　p.80
　5.4　Passing Maneuvers　p.82
　5.4.1　Speed during Passing Maneuvers　p.84
　5.4.2　Lateral Distance during Passing Maneuvers　p.85
　5.4.3　Longitudinal Distance during Passing Maneuvers　p.87
　5.5　Pair Riding Maneuvers　p.88
　5.6　Summary　p.90

6.The Signalized Intersection Analyses
　6.1　Dynamic Motorcycle's Lane　p.92
　6.2　Longitudinal Threshold Distance of a Motorcycle　p.93
　6.2.1　Longitudinal Threshold Distance 1　p.93
　6.2.2　Longitudinal Threshold Distance 2　p.94
　6.3　Deceleration Models　p.96
　6.3.1　Free-deceleration Model　p.96
　6.3.2　Following-deceleration Model　p.98
　6.4　Acceleration Models　p.102
　6.4.1　Free-acceleration Model　p.102
　6.4.2　Following-acceleration Model　p.103
　6.5　Reaction Time Distribution　p.107
　6.6　Maneuver Model　p.108
　6.6.1　Lane Selection Model　p.108
　6.6.2　Gap Acceptance Model　p.109
　6.6.3　Performance of the Maneuver Model　p.110
　6.6.4　Adjacent-lane-acceleration Model　p.111
　6.6.4.1　Free-acceleration Model　p.112
　6.6.4.2　Following Models　p.112
　6.6.4.3　Adjacent-gap-acceleration Model　p.113
　6.7　Performance of the Models　p.117
　6.7.1　Performance of Acceleration/Deceleration Models at the Green Time　p.117
　6.7.2　Performance of Acceleration/Deceleration Models at the Red Time　p.119
　6.8　Summary　p.120

7.Conclusions
　7.1　Research Summary　p.122
　7.1.1　Midblock Analyses　p.122
　7.1.2　Intersection Analyses　p.123
　7.2　Further Studies　p.126

　References　p.127

　Many urban areas in the world are confronted with transportation related matters. In well developed countries, these often involve four-wheel vehicles, whereas in developing countries, transportation problems involving small size vehicles such as motorcycles or moped are more common. However, very little empirical research has been conducted regarding the traffic operation of motorcycles. This dissertation addresses the comprehensive analysis and the development of microscopic models of motorcycle traffic at links and at signalized intersections through videotaping technique.
　For midblock analysis, a method for estimation of motorcycle units （MCUs） is developed. The method is developed with a consideration of dynamic characteristics of moving vehicles. It expresses the correlation about speeds and occupied spaces between motorcycles and other types of vehicles. Moreover, statistical analyses of empirical data are utilized to develop the speed distribution and the headway distribution of motorcycle traffic. Data analysis selection consists of four sections, in which two are dedicated exclusively for motorcycle flow, and two other sections for mixed traffic including non-motorized and four-wheel vehicles.
　The characteristics of passing and paired riding maneuvers of motorcycles on exclusive motorcycle lane and undivided roadway are represented. The passing characteristics in this study are analyzed upon （i） individual speeds and speed differences between passing and passed motorcycles; （ii） lateral spacing of passing and passed motorcycles; （iii） longitudinal distances from beginning and ending of passing events. Statistical data reveals that facility types, exclusive and undivided roadway, significantly influenced not only relatively different speed and lateral distance but also both longitudinal distances before and after passing events. The paired riding maneuvers are described by statistical analyses of speed and lateral spacing. The threshold for paired riding including speed difference is also determined.
　For intersection analysis, a model framework of motorcycles' behaviors at signalized intersections is proposed. Since motorcycles are much more flexible than four-wheelers, the adapted definition of a motorcycle's lane has been introduced to identify the proper leader of a given motorcycle. The dynamic motorcycle's lane is not stable on a roadway as a normal lane but flexible according to the subject motorcycle's position. The width of motorcycle's lane can be defined as the width of the operating zone of the subject motorcycle. Then, motorcycles' behaviors are treated similarly to cars' by applying acceleration/ deceleration and lane changing models with some modifications. Two regimes in the acceleration/deceleration models are proposed in the study: the free regime and the following regime. At every observed interval, the particular regime is specified by a longitudinal threshold distance. That threshold is estimated upon the kinematics concept, the function of speed and motorcycles' characteristics. If the distance headway between the subject and leading vehicle is larger than the longitudinal threshold distance, the rider is assumed to be in the free regime. Otherwise, the rider is in following regime. The distribution of reaction time for motorcycle population is also developed. The longitudinal threshold distance and reaction time capture the variations over motorcyclists.
　The maneuver model is developed in order to model when, where and how motorcycles maneuver in queues at signalized intersections. The threshold distance is estimated to identify when motorcycles need to maneuver. The maneuver process used to determine where motorcycles maneuver consists of two parts （i） lane selection model and （ii） gap acceptant model. In the first part, a motorcyclist faces three options, traveling in the current direction or change to either left or right direction. This procedure is repeated at every timed interval and modeled by a logit model. After a desired lane is chosen, in the second part, the motorcyclist looks for an acceptable gap which is modeled by the gap acceptance model. The available gap is accepted if it is greater than the critical gap, which is a function of the different speed between the subject motorcycle and the leader. All parameters of the lane selection model and gap acceptance models are jointly estimated in order to explain the completed behavior of motorcycles in maneuver. The adjacent lane acceleration model has been developed to estimate how motorcycles accelerate/decelerate during maneuver.
　The adjacent lane acceleration model is classified into three regimes （i）［ free-acceleration regime, （ii） following regime and （iii） adjacent gap regime. The lead and lag threshold distances determine what regime motorcycles belong to.
　The parameters of driving behavior models are estimated using microscopic motorcycle data collected from video images in Hanoi and Hochiminh city, Vietnam. The motorcycles' positions regarding time events were calculated according to screen coordinates then converted into roadway coordinates by using SEV software, developed in the traffic lab for specific purposes. Validation results of the proposed models using the comparison between actual and estimated trajectory data are also presented.

　本論文は，「Analysis of Motorcycle Behavior at Midblocks and Signalized Intersections（単路部と信号交差点における自動二輪車の挙動分析）」と題し，7章より構成されている。
　第1章「INTRODUCTION」では，既存の自動二輪車の挙動モデルの持つ問題点について述べ，本研究の必要性と適用範囲を記述している．
　第2章「Literature Review」では，既存の自動二輪車の挙動分析に関する論文や，乗用車の追従理論に関する論文のレビューを行っている．
　第3章「Methodology」では，動的自動二輪車レーンの考え方や，減速モデル，加速モデル追い越しモデルの概念を説明している．
　第4章「Data Collection and Analysis」では，モデル分析に必要となるデータの取得方法と，そのために開発したソフトウエアーの説明を行っている．
　第5章「The Midblock Analyses」では，単路部において，バス，乗用車，自転車の自動二輪車等量を求めている．また，自動二輪車の車頭時間や速度の分布の特性について考察を加えている．
　第6章「The Signalized Intersection Analyses」では，信号交差点における自動二輪車の減速，加速，追い越し，レーン変更，ギャップアクセプタンス等のモデルの同定を行うと共に，モデルの再現性の確認を行った．
　第7章「CONCLUSIONS」では，各章で示したモデルの特徴を概観し，今後の課題をまとめている．
　以上のように，本論文は，今までほとんど分析の行われていなかった自動二輪車の挙動をモデル化することにより，新しい知見を与えている．よって，本論文は工学上及び工業上貢献するところが大きく，博士（工学）の学位論文として十分な価値を有するものと認める。