東京大学大学院工学系研究科 都市生活学・ネットワーク行動学研究グループ

In this doctoral dissertation, three significant design attributes are identified for Bus Transit Route Network Design Problem (BTRNDP) for developing countries i.e., multi-dimensional transit vehicles, travel time reliability and elastic passenger demand which shouldn’t be neglected while solving BTRNDP for developing countries. First, a bi-level optimization model is exercised to solve the design problem of exclusive bus lane arrangement in a multi-modal transportation network. Then based upon the identified design attributes and the concept of exclusive bus lane, a reliability-based multi-dimensional transit priority system including small commercial vehicles is proposed for developing countries to reduce the on-board reliable travel time cost of all the travellers. A tri-level optimization model considering variable transit passenger demand under a certain fixed total demand is exercised to achieve defined objectives of the study. Three objectives at each level are defined including minimization of the reliable travel time of public bus passengers, maximization of the passenger demand for private commercial vehicles and minimization of the reliable travel time of all travellers, respectively at upper, middle, and lower levels under different operational and resource constraints. An iterative analytical approach is adopted with UE assignment procedure for motorists and headway-based transit assignment procedure for passengers to solve multi-dimensional BTRNDP for developing countries. The results indicate that the proposed model can effectively achieve the defined optimization goals at each level. A multi-dimensional transit system can effectively be used in cities of Pakistan and other developing countries where the different size of public and private commercial vehicles exists.

Landownership behavior is a recursive behavior in itself, and at the same time, it is another recursive behavior in terms of being interactive with other economic agents. In this doctoral dissertation, I formulated a long-term, dynamic, and reciprocal landownership behavior model to reproduce the recursive nature of landowners' landownership behavior, and demonstrated landownership behavior over a period of nearly 130 years from the modern era to the present. The model takes into account the heterogeneity of landowners, and at the same time, by sampling the state space based on landowners' cognition, it is possible to evaluate the behavior quantitatively, which could not be achieved by the previous historical approach. In addition, the simultaneous estimation of land transaction behavior and pedestrian behavior enables us to evaluate the interaction between land and traffic on a micro-scale.

In response to the shift in the central theme of urban administration from "maintenance" to "utilization" and "management," this study aims to develop a detailed analysis method of urban activities, especially private activities, and to apply it to administrative practice in order to improve the sophistication of urban management.

Translated with www.DeepL.com/Translator (free version)

For this purpose, we have developed a low-cost and simple method for estimating detailed OD tables by mode of transportation between zones by combining PT survey data and transportation-related big data, as well as a pattern analysis based on a behavioral survey conducted on the elderly, who have a particularly high proportion of private activities. Furthermore, we developed a policy evaluation method for private activities in areas centering on railroad stations, making effective use of existing data held by local governments, such as basic survey data and PT survey data. As a result, we were able to obtain a detailed understanding of private business activities for which data were not available from other statistical surveys such as the National Census, and which had been difficult to obtain with high accuracy using methods based on previous research.

In addition, based on these results, the future of urban planning and urban transportation surveys in administrative practice is summarized.

This paper proposes a public transportation network plan, service routes and management method of on-demand buses in the damaged area after the Great East Japan Earthquake, and analyzes its application in an actual area. To support the operation of demand transportation with a structured programming method, we proposed a fast algorithm for the optimization of transportation routes due to daily changes in reservation. Next, as a solution algorithm for the optimization of public transportation network, we implemented an indexing method that can efficiently handle a huge set of route options using a structured programming method, and applied the CE method as a solution method, and evaluated its performance as a computational method with extensibility to user equilibrium allocation. Finally, since it is often difficult to actualize the demand of public transportation in the reconstruction of the damaged area due to various circumstances, we evaluate the public transportation policy using latent demand indexes, and proposed a network evaluation method to quantify latent demand in the network planning area using the data envelopment analysis. As a result, we found that the construction and relocation of the upland complexes allowed us to distinguish between inefficient areas and areas where demand was manifesting itself, as well as to show analytically the expansion of the frontier.

The Markovian path choice model is an approach that avoids the problem of path enumeration, which is its biggest problem, by evaluating the path selection probability as a product of the transition probabilities between links, which are elements of the path. In this study, we attempted to develop a framework for path selection analysis based on a Markovian path selection model, which is based on 1) a path observation and estimation method focusing on the link specificity of observation error variance, 2) the introduction of a spatial discount rate for generalization of sequential path selection models, and 3) proposed a network description and constraint method to solve the computational instability caused by the existence of infinite cyclic pathways, and attempted to solve the computational problems of Markovian path choice analysis, as well as to expand a consistent analytical framework for observation, modeling, estimation and allocation to space-time network, and thus enable the proposed system to compute the problem of schedule network design for pedestrians on a 1 km square scale, which has been difficult to solve in the past.

In this study, we focus on evacuation from a disaster with a short grace period, and study how to model and control evacuation behavior based on the real situation. Three factors were considered in the decision to start evacuation: future risk, cooperative behavior with others, and the impact of evacuation of others. A dynamic discrete choice model with relaxed complete rationality was introduced to assess future risk, and the MPEC-type algorithm was used to solve the problem. We also studied the effects of spatial control and leading evacuation by focusing on joint evacuation. The joint evacuation has a positive effect of helping others and a negative effect of delayed evacuation due to support, and optimal measures were discussed based on the positive and negative impacts. We also evaluated a spatial control strategy to minimize the time of completion of evacuation applying the maximum principle.

This study focuses on the historical transition of survey methods in modern urban planning, and by scrutinizing the process of promoting urban planning projects through local councils and committees, we have summarized the history of the survey as a survey history and organized the issues of urban planning surveys today. Next, based on these issues, we developed a new algorithm for identifying transportation systems as a probe-person survey technique using the SVM to measure diversified living areas for the formation of compact city structure and to make the survey more detailed and automated. Finally, we conducted an empirical analysis of outcomes of policies for the development of intensive urban structures, and proposed a method for future urban planning surveys.

In this study, we propose a method of quantitatively assessing the effectiveness of a policy in developing a plan for urban revitalization. Using a normative model based on simple behavioral principles and a path selection model and a multiple destination selection model based on detailed observational data, we proposed a framework for multi-dimensional and multi-layered analysis and evaluation of both historical time change and current behavioral characteristics.

In this study, we treat individual subjects, i.e., buildings and sites, as components of the city from a microcosmic point of view, and try to describe the individual changes in the renewal process of existing residential areas by modeling. Based on case analysis, we aim to understand the limitations of micro sociological research on the usefulness and generality of implications, and to understand the mechanism of spatial formation focusing on the process of renewal and interaction of existing residential areas.

This study is based on the view that it is necessary to have a positive paradigm shift in the future urban structure in the era of declining population, and that the review of city planning roads should be regarded as "one of the positive time management methods for the realization of future city structure". This paper discusses the concept of time management in city planning for the realization of future city structures, focusing on the important concept of time management in the field of city planning.

There is a lot of interest in mechanisms design that autonomously lead to the desired social state by providing incentives to each actor because of the weakness of urban planning impacts. In the field of transportation planning, for the efficient operation of a shared-mobility service , which has recently been attracting much attention for its potential introduction, we propose a service auction and study the design of a mechanism to maximize social welfare. In order to satisfy the spatio-temporal OD connectivity and autonomous efficient operation of mobility sharing, it is necessary to design the auction mechanism of tradable permit mechanism and clarify the nature of its auction. Therefore, in this study, we design a service auction market with desirable characteristics and conduct an empirical experiment of the service auction, and investigate the applicability of the service auction of mobility-sharing from both theoretical and empirical perspectives.

In recent years, the importance of street space connecting transportation nodes has been increasing due to the diversification of transportation modes, such as the emergence of self-driving cars, as well as the reevaluation of pedestrian space. In evaluating street spaces, route choice models that can evaluate the utility of street spaces are useful, but in order to capture the trade-off relationship of convenience between transportation modes in street space design, it is necessary to describe the relationship between the spatial structure of streets and the interaction between transportation modes. In this study, we proposed a method that can efficiently estimate stable equilibrium states in the presence of endogenous interactions by using an adversarial inverse reinforcement learning method that is consistent with conventional models from a game-theoretic viewpoint. We also proposed a framework that can obtain suggestions for street space design from a route selection model by combining a link feature extraction method using satellite images and a microscopic route selection model based on camera data, and showed that estimation using real data is feasible.

For traffic networks, we present a method to find the optimal control input in terms of energy and a method to evaluate the controllability of traffic networks and control systems by using controllability Gramians. The proposed method is applied to pedestrian and automobile networks to find the optimal control inputs and to solve the network controllability maximization problem using controllability indices. It was shown that when the sum of the diagonal components of the controllability Gramian and the determinant are used as indices to determine the control points, the target state can be reached with less energy than in the case where the indices are not used. The applicability of the method to real-world situations was evaluated using microsimulations, and it was found that control toward the target state was achieved.

Since the spatiotemporal distribution of risk sources, elevation, and other factors affect behavior during disasters, it is assumed that the heterogeneity of behavior between regions and disasters will be greater than in normal times. In addition, verification is neglected in the behavioral modeling field, which may suggest inappropriate planning. This study deals with destination selection in tsunami evacuation with a view to applying the model to hard planning such as resource allocation and network reinforcement, and analyzed 10 cities hit by tsunami in the Great East Japan Earthquake. In addition to the traditional MNL model, the IAP model, which considers a set of alternatives, and the SCL model, which considers spatial correlation, were constructed as models that describe the evacuation process in more detail. We also used machine learning models, which are expected to have high generalization performance in recent years, to validate the four models and clarify which model should be used for relocating to other areas. Furthermore, we classified cities by evacuation similarity and confirmed that relocation performance is enhanced by using data from similar cities.

In evacuation under uncertain conditions caused by conflicts, major disasters, and climate change, households and communities have been dispersed. This study attempts to establish the foundation of the theory of humanitarian return management in reconstruction and to demonstrate the theory by conducting a qualitative analysis of the process of reshaping gatherings in the areas affected by the Fukushima nuclear power plant disaster and by introducing the quantitative model of gathering participation, the dynamic habitat selection model, and the network design problem into the mixed research method. The results of the study show that the humanitarian return management theory can be applied to the process of recovery from crisis. The results revealed the existence of distance resistance that hinders women's participation in the process of reconstruction from the crisis, and suggested the possibility of macro-scale migration management (return / settlement choice) by supporting micro-scale migration (participation in gatherings). The study suggested the possibility of macro-scale migration management by supporting micro-scale migration (participation in gatherings).

In recent years, large-scale development has been carried out one after another, but there is still no indicator to evaluate the impact on the surrounding public space and existing resources. It is needless to say that it is important to clarify the long-term causal effects of development on the liveliness of a town, but it is not easy to quantitatively describe such effects, which fluctuate in a complex spatio-temporal manner. In this study, based on the idea that the essence of behavior principles and activities in real urban space is time consumption, we proposed a new TS-DID model that can express the spatio-temporal decay and growth processes of the effects of multiple interventions between time points from difference equations that introduce spatial autocorrelation and its spillover effects. The proposed model is applied to Shibuya, where a large-scale redevelopment project is underway, using PP data from multiple time points, and it is revealed that development facilities located in clusters mutually compete for the consumption time of pedestrians in the surrounding area, resulting in the localization of time consumption space by increasing concentration in a specific core and sparsification in the surrounding area.

Bayesian Network (BN) can model data-driven dependencies among variables by estimating the graph structure of variables. In the field of transportation, BNs can explicitly reflect the changing decision-making structure of individuals in the form of variable graphs, depending on the data, and are being applied to activity-based models (ABMs). This study aims to improve the usability of BN models as ABMs based on the constructed variable graphs by introducing prior knowledge of the transportation field into BNs and stabilizing structural learning. First, we introduce an Object-Oriented Bayesian Network (OOBN) to ABM and propose a method for structural learning and activity generation that takes advantage of prior knowledge of the transportation field. Next, we propose Graph Averaging to stabilize learning by selecting graph substructures that enhance the explanatory power of the BN model through a deep generative model. Finally, we demonstrate the effectiveness of the proposed method through numerical experiments assuming the true model and validation using real data from Tokyo PT.

The purpose of this paper is to compare and reference efficient urban policies by aggregating a large number of existing urban evaluation indicators using statistical methods. In past urban evaluation, researchers and administrative agencies selected indicators arbitrarily, which may have resulted in bias in the selection of indicators and evaluation axes. In particular, the Ministry of Land, Infrastructure, Transport and Tourism's "Handbook for Urban Structure Evaluation" proposed more than 40 indicators, but these included many indicators with strong correlations, which could lead to inefficient city evaluation. In this study, we attempted to extract efficient urban evaluation indicators by performing correlation and factor analysis based on data from 42 cities. Using this new evaluation method, we also make comparisons among cities and over time, and propose indicators that are effective for monitoring urban policies.

In this study, we attempted to construct a railroad congestion control method focusing on the following three points: First, we considered travelers' stay in station areas and handled congestion in station areas in addition to trains. The second is to theorize a billing control method with a point system. The third point is to make it possible to apply the system even when the total demand between ODs is unknown. This was achieved by introducing day-to-day dynamics related to travelers' activity choices, etc., and replacing the objective function with a control problem that matches the Lyapunov function. In addition, numerical experiments simulating the Yamanote Line service suspension in January 2023 showed that demand control using the station space in the outer edge of the city is effective when the network in the city center is disrupted, and that the effect of the demand control is enhanced by the point allocation.

In recent years, much attention has been paid to the development of streets and plaza spaces centered on pedestrian circulation in urban redevelopment. Particularly around terminal stations in Tokyo, new pedestrian networks and plazas have been formed in response to the opening of new train lines and the renewal of station buildings. However, when private operators pursue short-term profits, the attractiveness of the entire city is not necessarily enhanced. This study focuses on Shibuya, where a large-scale redevelopment project is underway, and analyzes the impact of the redevelopment on land ownership and land rents. Specifically, we propose a two-stage model that combines pedestrian circulation behavior and dynamic operation of land, and use the model to evaluate policies. First, a model of pedestrian circulation behavior and land operation is constructed, and second, the model is estimated using real data. From the analysis results, suggestions for redevelopment policy and land operation optimization are obtained.

Currently, the periodicity of the disaster period and the impact of the policy on the future distribution of residential areas are not taken into account in advance reconstruction. In this study, we modeled a sequential turn-by-turn process in which residents' choice of residential area and the government's choice of pre-disaster reconstruction policy interact with each other, and developed a sequential evaluation method for the optimal policy. The choice between the residents and the government, which is based on future utility and investment benefits, is interdependent, making it difficult to calculate the probability of choice analytically. Therefore, the least-squares Monte Carlo method of option theory was introduced to specify a regression function to estimate future utility and benefit. We assumed that the pre-disaster reconstruction would consist of soft measures, such as inducing residential relocation through disaster insurance, and hard measures, such as the construction of breakwaters and seawalls. The relocation of settlements is a long-term, low-cost measure for disaster prevention due to the economy of agglomeration. On the other hand, the construction of hard measures has an immediate effect, but the timing of construction needs to be optimized in consideration of accumulated maintenance costs and age-related deterioration, and may create a moral hazard by giving residents in coastal areas an excessive sense of security and discouraging inland relocations. The optimal policy at each point in time should be judged comprehensively based on the topographical characteristics and residential conditions of the city.

This study proposes a framework for optimal control of staged evacuation, which controls the start time and destination of evacuation for evacuees in the event of a heavy rainfall disaster. First, parameters representing learning effects and forgetting were introduced into a dynamic evacuation destination selection model for heavy rainfall disasters to describe a dynamic learning process in which evacuation behavior changes as cognition is updated by disaster risk information. Using this model of evacuation behavior, an optimal control model for staged evacuation was developed to determine when and to which area evacuation information should be disseminated. The model is supported by a predictive control framework that allows the plan to be modified in response to changing conditions. A zone-based evacuation traffic simulation was developed to achieve the fast predictive computation required for predictive control. The staged evacuation framework worked well when congestion levels were low and evacuation grace times were long, but when congestion was extremely bad and evacuation grace times were short, the effectiveness of staged evacuation could not be confirmed. The results suggest the need for combined traffic supply control and demand control under highly congested networks.

We model dynamic residential choices after a large-scale disaster using the DRL model. The time discount rate reflects the uncertainty of the disaster and the prospects of reconstruction projects. The impact of others' choices on each household's residential choice is defined as an externality term, which is estimated endogenously by structural estimation. We present a network design framework to determine the Spatio-temporal allocation of rent subsidies. The objective functions are maximizing the total surplus and maximizing the population in the affected area. The Pareto solution is not reached by conventional uniform support or independent recovery plans conducted by each municipality, and Braess's paradox may occur as shown in the theoretical model. Therefore, it is meaningful to take a long-term perspective and manage the reconstruction plan in time and space for a wide area.

In this study, we attempt to understand and model the mechanism of people's self-restraint behavior under the Covid-19 epidemic. The results show that people choose destinations in the surrounding areas of their homes, while leaving out densely populated areas such as the city center. In addition, the model was extended to a fully dynamic destination selection model by using a two-stage model that takes into account the formation of choice sets, and by fitting it to a learning algorithm for the bandit problem, in which only the utility of the selected destination is observed and learned. Real data were estimated, and the parameter value for population density decreased even after the declaration of the state of emergency was lifted, while the parameter value for the number of stores recovered after the lifting of the state of emergency.

Based on the CUE model, this thesis proposes a new location equilibrium model that aims to introduce the impact of remote work and to refine the description of the labor market. The previous studies assumed that households commute to their firms. On the other hand, people's lifestyles are changing rapidly due to the corona pandemic. This study will develop a location model that can respond to these changes. In addition, since the impact of remoteness is significant, this study attempts to represent the employment relationship between firms and households in a remote-diffusion society using matching theory. It also proposes a framework for integrating the location model and the matching model. This study also tries numerical experiments. It analyzes the location patterns and discusses the differences between remotely and non-remotely firms and households. Finally, I apply the matching model to the equilibrium location distribution to check the distribution of unemployed people.

In this study, we digitized historical records of land ownership and use over a long period. Then, we analyzed the combination of land ownership and use, building type and height, and showed the metabolism mechanism of land. First, we showed the significance and methods of digitizing the data. Next, using cadastral data, we showed the structure of land ownership, such as individual / corporate and old / new landowners. After that, we analyzed the building use and height for each landowner attribute, showing the relationship between land ownership and use. There were many individual-to-individual, individual-to-corporate, and corporate-to-corporate ownership transfers before 1960s, 1970s-80s, and after 1990s, respectively, and building use and height changed significantly in the 1980s. Finally, by focusing on the characteristic landowners and urban tissue, the formation process of the districts and spaces was clarified along with the intentions and strategies of the landowners.

In this study, an optimization problem for the operation of passenger-and-freight vehicles is described as a combinatorial optimization problem of the travel demand of passengers and freight in a sparse regional transportation and logistics network. A fast matching algorithm for recalculating feasible routes in response to sudden changes in schedule is constructed by route enumeration and indexing, taking into account the uncertainty of demand. This study also conducted numerical experiments on several scenarios regarding heterogeneity in the trip patterns of passengers and freight, focusing on their round-trip and one-way trips especially. For the First-Come-First-Served strategy, it is confirmed that the round-trip request deteriorates the matching efficiency in terms of its success rate and total travel distance. Using all the enumerated and indexed routes, the marginal contribution to the matching candidate is evaluated by Shapley value, and its effectiveness for the assignment problem of vehicles to customer sets is demonstrated.

In this study, we focus on the fact that we can regard vacant lands as land waiting to be used and first propose a model that can represent the formation of vacant land in real-time, incorporating queuing theory. We find a nonlinearity in the progression of vacant land creation and differences in response to the number of vacant spaces and vacant time. Then, we propose a model in which buildings, parking lots, and vacant lots form a spatiotemporal pattern through a spatial correlation based on behavioral theory. We show that the distribution of vacant lots and parking lots undergoes a phase transition. Both models are validated using actual data, and we confirm that the actual city is close to the critical state. With the evaluation of response time to parameter variation, we confirm that the diversity of the urban form is due to the high criticality rather than parameter variation.

In this study, we develop a new land use and transportation model that takes into account the mutual influence of households and firms using a bilevel optimization framework. It enables us to analyze various optimal policy evaluations over time. We define a multi-objective optimization problem assuming both total revenue maximization and household utility maximization, and show the optimal supply strategy and location situation within the urban area corresponding to each behavioral principle, confirming the usefulness of the proposed model. Furthermore, we apply the model to the Tokyo metropolitan area for the last 30 years, and it is confirmed that the tendency to emphasize business convenience in the location selection of companies and the sensitivity of business floor space rent to the density of employees increase accordingly. In addition, a time series analysis of location and transportation data suggests that there is a spatial bias in land development and transportation development.

This paper attempts to construct a method for simultaneous estimation of Origin-Destination traffic distribution and route choice parameters, focusing on pedestrian behavior. For the objective function, we propose an estimation algorithm that guarantees the uniqueness of the converged solution from the viewpoint of information geometry. And we attempt to optimize the park development plan from a quantitative viewpoint, such as the expected utility of tourists. The optimization problem is huge computational cost because the solution set is discrete and the assignment is computed by recursive logit model. Therefore, we propose to solve the assignment calculation by training a feedforward neural network using the estimated model and supervised data of the network and objective function values.

In this paper, we tackled the problem of missing data of fatalities in evacuation data and present a framework for analysis and evaluation of pre-reconstruction planning. First, the activity path choice during evacuation is formulated by the Discounted Recursive Logit model, which represents dynamic decision making. Then, we proposed a cost-sensitive maximum likelihood method to correct for survivor bias due to the missing data of fatalities. Applying the proposed model to the data of the Great East Japan Earthquake, it was shown that ignoring survivor bias will lead to misunderstanding evacuees' behavior. In addition, the framework of the evacuation network design problem is presented to find the Pareto solution set for two objective functions: maximizing the number of successful evacuees and minimizing the cost of road capacity expansion. The model is applied to Yawatahama City, Ehime Prefecture, and the effects of relocation to higher ground and road construction are discussed.

In this paper, we introduced an interaction term into the discrete-continuous model and constructed a model that correlates discrete activity choices influenced by the activity choices of others, with continuous time allocation choices. To describe social network constraints on discrete choice set, we extend the solution of the stable marriage problem by the Gale-Shapley algorithm as a bigamy matching pair generation problem. We applied these frameworks to the schedule matching of people requiring assistance and their supporters in the community ,considering the concept of capability. By adopting the social network as a choice constraint in the scheduling model, we have developed a model of district-level daily sharing services based on intimate relations and the finer transportation system for a growing number of people who have difficulty getting out of their house in an aging society.

We propose here machine learning algorithms based data-oriented sampling protocols that sequentially generates the destination alternatives at each choice stage explicitly without causing any mathematical complications in the model. Two type of sampling protocols are proposed here: rule based sampling and cluster based sampling. The number and type of destinations that are used as alternatives have huge impact on the stability of parameters and model performance. The stability of parameters and model performance are evaluated with proposed sampling protocols under different scenarios, and analyzed in comparison with conventional random sampling protocol in detail. The proposed methodology would potentially explore the understudied aspect of destination choice modeling, and provides deeper understanding about the full-day decision making dynamics of the decision makers. The results are expected to be used for the formulation of time-dependent origin-destination table in the study area that eventually assist the planners in policy design and operation.

In this study, we analyze the relationship between the past data of containerized transport and corresponding infrastructure investment overtime.The analysis has been divided into two parts: historical analysis of global containerized trade and local intermodal container traffic assignment on Pakistan’s hinterland network. Using the two Layered intermodal network assignment model, the global containerized transport demand has been allocated to the hinterland network of Pakistan. The result shows that applying Dedicated Freight Corridor (DFC), which is a railway corridor exclusively used for the freight transport from sea ports to inland economic hubs, can improve network effectiveness in assigning container traffic and the goals of increasing the exports of Pakistan presented in Pakistan Vision 2025 can be efficiently achieved by the proposed policy.

This paper focuses on the existence of landowners in the city and aims to clarify the mechanism of urban transformation in hot spring areas up to modern times, using both a conventional bibliohistorical approach and a statistical approach based on non-aggregate analysis of landowners' behavioral norms. Digitizing and partially restoring the previous land register book, we analyzed land ownership history of landowners combining with primary historical data and revealed the process of renewal of regional infrastructure and the land ownership structure where oligopoly and social mobility of some landowners exist simultaneously. We formulated a non-aggregate model that takes into account the latent heterogeneity among landowners, which is a selection problem of land ownership type, and employed the EM algorithm to enable parameter estimation. These results indicate that changes in land ownership trends due to infrastructure renewal may have contributed to different urban formation processes among eras.

We aim to explore the mechanism of the residential area expansion, which is in high risk of natural disaster, focusing on the development of the transportation infrastructure and city planning after modernization, in Sanriku region and Hijikawa region. It is revealed that the direction of urbanization was highly influenced by the transportation infrastructure such as train stations and highways. Some of the cases are insightful enough to show the possibility of optimizing the location of residential areas by infrastructures, for example, using high risk area as industrial zone in Kamaishi, and displacing a highway to the safe upland in Ofunato.

The estimation of behavior model based on non-aggregate analysis of pedestrian behavior is required in order to analyze pedestrian behavior in space of about 1km square with high accuracy. In this study, we propose the estimation of behavior model from fixed observation acquired by video footage and movement observation for each person acquired by GPS and Wi-Fi. Previous methods based on movement observation require route data as input. In contrary, we define the estimation of behavior model from movement observation based on entropy and propose the estimation based on cross-entropy as the estimation of behavior model from fixed observation. This method is applied for simulation data by twin experiment. As a result, we show that this method has validity against previous estimation methods and the estimation of entropy has robustness against observation error.

In this study, we describe the formation process of road networks in the poor and densely populated urban areas in Japan and France, which are known as "barrack towns" (Tokyo and Osaka) and "insalubrious zones" (Paris), from a historical perspective. While these districts have been the subject of inner-city studies, the formation of the road network has only been discussed in the context of the national policy on road and housing clearance in the 1970s. This paper describes the history of the formation of the road network, its legal and budgetary background, and its historical background. Target era is from when it was established as a modern city to 1960s, and target district is for mer "Koishikawa ward" in Tokyo, the district which is now defined as "Naniwa ward" in Osaka, and "the 19th arrondissement" in Paris. The results revealed that these three districts are located on the periphery of modern cities, and they naturally formed dense urban areas due to the strong financial constraints on the city's road construction, but few areal improvement projects took place due to expropriation problem whether the project was to be initiated by private sector or public sector. It was necessary to wait for policy intervention by the state for the improvement projects to take place in those areas.

Shopping centers (SCs) have been regarded as a primary factor of decline of the city center, but they play an essential role in the city, such as shopping and entertainment, and have been evolving day by day in response to the transformation of the city. The purpose of this study is to continuously analyze the trends in the location, opening and withdrawal of SCs from the 1970s to the present in three prefectures in the Kanto region (Tochigi, Gunma and Saitama Prefectures), to obtain implications for more strategic use of these shopping centers in urban planning. The analysis confirms that facilities are becoming larger and more suburban, and showed that the number of SCs withdrawn in recent years has been increasing, and that there are certain geographical and facility characteristics of the shops withdrawn. Based on the trend and actual cases regarding diversification of the functions of SCs such as food and drink, services, entertainment, culture, public service and welfare, it has been revealed that there is a gradual difference in the diffusion rate of each function, and that the location’s environment of the SCs have a significant impact especially on the success of the introduction of the public function of the SCs.

The purpose of this study is to investigate the process of transformation of the transportation network and regional structure through extrapolation of the modern transportation system in the regions where the transportation system was based on riverboat traffic. This study was conducted in Okayama Prefecture and the vicinity of Okayama City, mainly based on the literature survey. This study set the target field in Okayama Prefecture, where the boat transportation network was formed on the Yoshii, Asahikawa and Takahashi Rivers, and revealed that the river transportation network was formed in connection with the railroad as a result of the gradual improvement of the railroad and thus the nodal city rose and fell, and that the former regional structure along the rivers from the upstream to the downstream was transformed by the construction of railroad. Also this study showed that, in the vicinity of Okayama City, which had been the hub of river transportation of Asahikawa River, river transportation declined and revived due to multiple factors such as sedimentation in the Asahikawa River, the construction of a harbor railroad, and improvement work in the Asahikawa River, where the shipping lanes and loading docks were constructed, and that the Port of Kyobashi in the center of the city prospered as a node between the city and shipping, but gradually the center of the port and manufacturing industry moved to the seaside and the structure of the surrounding area was transformed.

In this study, in order to quantitatively analyze the phenomena at the highway merging sections and its spillover effects on traffic flow, we constructed a simulation model extending the simulation of the multi-lane cell transmission model, and evaluated the phenomena at the highway merging sections due to the spread of automated vehicles in the future. Assuming that the parameters representing the limits of human cognition and machine performance change with the spread of automated vehicles, we evaluated the traffic flow in the case where automated vehicles are commonly used. The results are as follows: 1) congestion of traffic flow will be seen suddenly when the minimum gap value related to the feasibility of lane change becomes small and lane change is frequently happens; 2) congestion will be relieved because the impatience to merge is relieved and the merging positions are dispersed within the merging section; 3) congestion will relieved when vehicles on the main lane are not affected by the merging vehicles and thus excessive vehicle deceleration is mitigated, and 4) congestion will be worsened when the duration of reaction delay is reduced, except when the minimum gap is short and thus aggressive lane changes occur.

This study deals with public transportation route optimization. In recent years, especially in rural areas, there is an urgent need to improve the convenience of public transport. In today's world, it is essential to plan a combination of multiple public transportation services, such as shared taxis and buses, and to evaluate the convenience of these services, taking into account the competition with trains and private cars. In addition, the demand that planners have originally assumed in public transportation plans often changes. Based on these premises, this study formulates a two-stage optimization problem: utility maximization theory for predicting user demand and maximizing convenience for planners' route decisions. Multiple types of public transportation services are described as capacity constraints in the route choice problem of the planner, and by incorporating the demand and availability of each mode of transportation that is estimated from trajectory data into the evaluation function in advance, an appropriate evaluation of convenience can be made based on the existence of competition. However, in the capacity constraint description of the route choice problem, a special network with information proportional to the fourth power of the real network is required. In this study, we have made it possible to solve the problem in a realistic computational time by contracting the network. Lastly, the proposed methods were applied to the project that opens commuter buses to the public in Kurobe City, which was currently underway, and its practicality was confirmed.

The purpose of this study is to summarize the history of the formation of expressways in Tokyo, Paris and London, and to discuss the appropriateness of the road network in Tokyo and Paris by measuring the social benefits. Specifically, this paper summarizes the road development plans of Tokyo, Paris and London, and examines the differences in the degree of realization of these plans in each country. We also compared systema and showed how, in France, private companies were entrusted with the construction and management of roads from the 1970s, while in Japan, with the aim of achieving balanced development of the country, the public corporations kept extending the repayment period endlessly and increased their debt by not adopting a line-by-line accounting system. Furthermore, through the evaluation of the social benefits of the hypothetically "possible" road network formation, the process of cross-sectional urban transportation network formation can be analyzed, and the mechanism of road network formation is re-evaluated by analyzing the system of road development behind the process. The results show that there is a difference in the way the Pareto frontier is formed between Tokyo and Paris, and pointed out that, while the design philosophy of infrastructure in Paris smoothly shifted from relieving congestion in the city center to the maximizing the effect of overall network, the design philosophy of infrastructure in Tokyo wavered, and the development of expressways in the Tokyo metropolitan area was a measure to the disordered growth of Tokyo to an extremely large metropolis.

The purpose of this paper is to understand the path choice behavior of pedestrians in railway station areas and the effect of spatial transformation on their path choice behavior, and we proposed an analyzing method for three-dimensional path selection behavior and evaluated real-world urban policies through simulations based on the simulation results.In the observation of traffic behavior, since the observation results used in the path selection model are stochastic, it is desirable to define the amount of information obtained by the observation, and it is very difficult to obtain true parameters in a path choice model that has a huge choice set and usually includes multiple variables. In this paper, we propose a probabilistic information estimation method based on geometric properties of the probability distribution of observed results, and a three-dimensional path identification method. We showed analytically and numerically that the determination of solution vectors based on probabilistic path selection results is important, and verified the effectiveness of this approach through estimation based on empirical data.

The purpose of this study is to propose a framework and method for redefining urban space as a history of the response process of urban organizations to the mediating nature of the multiscale space within a city and inter-cities. Firstly, based on historical research, this paper presents a framework for deciphering the relationship between multi-scale infrastructure, urban organization, and architectural composition from a number of materials. Next, we propose a method that applies network analysis to quantify the transformation of the urban organization within a city. Furthermore, based on the understanding of the transformation of urban organization, we propose to extrapolate an urban structure consisting of core and edges. Based on the proposed method, by organizing the constraints regarding infrastructure and pedestrian flow, we aim to orient the dilution of urban organizations due to ad hoc renewal, and present a method that serves as a basis for urban design theory to order the relationships among urban organizations.

In this study, we assume that urban typologies classified by population density, public transportation, and transportation issues in cities significantly contribute to the establishment of car sharing service, and based on these macroscopic indicators, we cross-sectionally evaluate cities where existing services are deployed to determine the feasibility of introducing car sharing service for the popularization of car sharing services. Firstly, we held case interviews to grasp the actual situation of services, and found that urban and social structures have a complex influence on the establishment of car sharing service. Based on this point of view, we assessed the feasibility of free-floating-type car sharing service, in consideration of urban typologies using a logit-type model, which enables us to compare the potential of existing services in the cities where they are deployed. In addition, we conducted a case study of one-way services demonstrated in Yokohama, Japan from a microscopic perspective, and showed that operating car stations can improve profitability.

In this study, we focus on the impact of heavy vehicles, especially overloaded vehicles, on the road maintenance of logistics enterprises. Road management operators can reduce the damage of overloaded vehicles on the roads by properly controlling them and guiding them to the high standard roads such as highways. Therefore, measures have been tried to install devices that automatically measure the vehicles and to designate the roads on which the vehicles drive and to monitor the compliance rate by using GPS technology. However, increasing the number of weighing devices to increase the accuracy of weight control requires high management cost. Moreover, depending on the way of road designation, logistics companies may have to make a large detour and the compliance rate may decrease. Therefore, we introduced a checkpoint policy that requires logistics enterprises to pass a certain number of nodes. The checkpoint policy reduces the number of weighing devices and indirectly controls the routes at the same time. The optimal arrangement of these checkpoints is calculated for the network in the Tokyo metropolitan area by using the cross entropy method.

The purpose of this study is to sort out the mechanism of the paradox regarding the car sharing system that when a new station is placed in a location with extremely unbalanced demand, the balanced equilibrium of car-distribution will be upset and result in lower revenues than before the addition of the station, and at the same time, to consider and evaluate the measures to improve the profit. At this time, we focused on the stochastic nature of the system's behavior, and formulated the profit maximization problem as a stochastic inventory management problem, and thus attempted to control it by optimal billing strategy. By using dynamic programming with Monte Carlo method in the solution algorithm, we have made it possible to compute even in systems with complex probability distribution of demand generation. Also in this study, we analyzed the detailed behavioral data obtained from the demonstration experiments and clarified the decision-making structure of service use. By building a simulation on the real network, we realized evaluating measures for real services.

We have used a demand forecasting system based on a person trip survey and a four-stage estimation method in conventional urban transportation planning. On the other hand, there have been remarkable improvements in the observation technology of human location information and in the performance of computers, and it is becoming possible to enumerate and compute a huge number of behavioral options on the network through long-term detailed personal observation. In this study, we have proposed a multi-scale simulation method. We assume that meso-micro scale policies such as the Olympic Games have macroscopic effects on the transportation system of the entire metropolitan area, and model the choice of travel and activity by using a reduced transportation network with cell-based division of the micro-rambling behavior corrected by PP data and the space of the entire metropolitan area centered.

Since people came to use automobiles, parking areas in the central city have been given in an unregulated manner in order to secure quantity. In this study, in order to consider the concept how we plan space in the central city, we first analyzed spatial factors of pedestrian behavior in the form of the occurrence of small activities and changes in time spent in the central city, using data from the Probe Person Survey (PP) conducted over the past eight times in Matsuyama City, Ehime Prefecture. We also proposed a design method for the spatial of pedestrians, focusing on the location of parking lots as an urban edge. We formulated and analyzed the balancing equations between parking characteristics (fees, capacity, etc.) and the inter-street environment in a simplified one-dimensional city using the parking distance to the destination as a parameter, and quantitatively calculated the unit of neighborhood formation.

In this study, we analyze the streets around stations as a network structure and identify the characteristics of the network, based on the idea that stations are the starting point of rambling behavior. The analysis targets not only the area around the station, but also compares it with facilities designed for circulation in order to highlight the characteristics of the station area. We focus on the closed network characteristics of the station-based network, calculate the number of routes, and propose it as a quantitative evaluation index of accessibility. Using these indices, the process of development of the network as seen from the historical changes around stations since the Meiji era, when railways began to operate, and evaluating future plans for station area development, we consider the future of the network around stations.

-

The purpose of this study is to analyze how people evacuate from the Great East Japan Earthquake in 2011, focusing on information acquisition, companions and transportation. This study focuses on the dynamic evacuation behavior of Rikuzentakata, which has not only a rias topography peculiar to Sanriku but also has a wide plain. This study aims to clarify the dynamic evacuation behavior during the 39 minutes between earthquake and tsunami, based on complex spatial and cord characteristics, and individual characteristics.

In the construction of a traffic control system, we need loop and sonic fixed-point observation, congestion estimation by data transmission, and time required for route estimation. In recent years, the development of GPS positioning technology has made it possible to obtain personal location data and acceleration with high accuracy. We can analyze the trajectory data by correlating the existing traffic control systems with the huge amount of fixed-point data. Based on this background, we develop a sensing method of traffic conditions for a virtual network and evaluate the proposed method for a real network.

Because the patterns of people's congregation are diversified, we need to change the conventional methodology of urban facility development, which is to get public land (or use existing vacant land) to improve the visual appearance and design a large box-like structure. In this regard, it will be important to have a better understanding of the relationship between people's actual movement behavior and their behavioral contexts. In this study, we examine how the spatial distribution of social networks works in various behavioral contexts and how it is connected to mobile behavior. We also collected social network and behavioral data on social networks by interviewing people. In order to examine the relationship between the behavioral context and actual migration, we focus on non-mandatory rambling behavior such as rambling, which is expected to increase in the future, rather than obligatory rambling such as commuting to work or school.

In this study, we proposed a traffic discrimination, an accurate pedestrian location estimation method, and a travel load model. Nowadays, it is easy to use GPS and accelerometers by smartphones, and these sensors can be used to predict what the user is doing. The model of transportation identification uses a support vector machine and hidden Markov model, which can discriminate between "walking", "bicycling", "car" and "staying" with high accuracy. The pedestrian position estimation is based on a rich set of acceleration data, which can compensate for measurement errors at a few meters. The travel load model is based on a discrete continuous model for the main purpose of movement and the amount of activity. The model evaluated urban space development policies.

We evaluated the impact of the streetscape on pedestrian route choice behavior.GPS technology allows us to understand walking behavior on the street. The study used a multinomial logit model to show (1) how pedestrians behave on the street and (2) how the configuration of the street affects pedestrian route choice. It was shown that sidewalk frontage length and sidewalk width were important for pedestrians, and that pedestrians tended to choose streets with a continuous street configuration similar to the one they had walked on before.

We proposed a concept and model of a one-to-one mobility sharing system based on game theory. Although there are many types of sharing systems, few theories and models take into account the theory of sharing. Although there are many players in a sharing system, we formulated it as a two-player game of one-to-one relationship, assuming a long-term relationship between two players and discussing the equilibrium state.

This study focused on the intensity of action and behaviour of vehicle traffic and decision-makers' interactions, and proposed a theoretical framework for the integrated assessment of transport systems in cities. This study revealed that highway users are more burdened than general road users with regard to parking fees and walking distance from the parking lot to the destination. Appropriate size of pedestrian rambling space and parking fees were obtained by formulating their urban stay and parking behavior. An empirical analysis of traffic behavior due to the effects of parking fees, parking location, highway and railroad fees was also conducted by static equilibrium allocation (SUE) using a real network in the Osaka metropolitan area.

We developed a data-oriented activity simulation model to construct a traffic control system based on a large amount of traffic behavior data. In particular, we proposed a method that enumerates activity type choice sets based on the location data of the probe survey. This method uses an algorithm that adds a sub-activity to the actual activity, and it is able to generate a set of alternative choices that individuals may choose compared to the previous method. This simulation model was able to evaluate the feasibility of introducing EV sharing in the Minato Mirai area of Yokohama and to understand the changes in mobility and activity patterns.

-

This paper focused on spontaneous street networks in European cities, such as Barcelona (Spain) and Siena and Bologna (Italy), in an attempt to extract the characteristics and principles of the pedestrian-oriented urban street and square system by analyzing the network according to the process of urban development. We found that skeletal streets established in the early stages of urban development coincide with the accumulation of the shortest routes, that central plazas, which function as nuclei or nodes, are located on highly mediated skeletal streets, and that plazas that connect multiple streets create a highly connected area of the street network.

This study involves the conservation and restoration of a redevelopment project in the historic center of the Old Town of Havana, the capital of Cuba. This district is a densely populated area with about 60,000 inhabitants in 2.14 km2 area and consists of a 1.2 km2 area within the old colonial wall and the surrounding area. Based on the field research, we will discuss the two-tiered structure of Old Havana and examine the practical program of the Conservation and Restoration Redevelopment Project by the Historical Office of the City of Havana. The spatial system in this project has been redefined as a spatial system of streets and squares, based on the Spanish Colonial Era's Plaza system. A pedestrian-oriented traffic axis connecting the squares is the center of the system, and the edge of the system up to the coastline is set up as a circulation space, and the ground floor of the building facing the street space is renovated as shops and a museum. By renovating the upper floors of the buildings while retaining their residential functions, the project has succeeded in creating a living environment for the street space. The historical office is the organization for planning and executing these programs and has created an autonomous form of operation that is linked to the social welfare program.

There are a lot of interests in the shared use of bicycles in the context of increasing accessibility and revitalization of urban areas, and we conducted an evaluation analysis of the measures to introduce this system. We implemented a shared-use bicycle system that allows users to ride and discard bicycles using IC cards and conducted a social experiment. In our experiments, we obtained multiple data: PP data that record detailed movement-activity data of individuals, SP data that express detailed LOS such as parking and bicycle parking fees, and choice data of the shared-use system. We analyzed the choice patterns of the system, and developed a behavioral model to represent the selection of tour patterns that represent individual mobility-activity patterns, and developed a method to evaluate the bicycle sharing system.

We tried to clarify the structure of dialogue in city planning by morphological analysis and homology analysis. The main perspective of this analysis was the semantic analysis of dialogues based on the meta-concept of faceted speech content, with text mining as a basic analysis method for linguistic data. By analyzing and comparing two types of data, interviews and community radio speech data, it was observed that the pattern of normative occurrence differed greatly depending on the interaction partner and that it was constant regardless of the interaction partner, indicating the necessity of an analytical frame based on a dynamic language perspective.

We tried to clarify the structure of dialogue in city planning by morphological analysis and homology analysis. The main perspective of this analysis was the semantic analysis of dialogues based on the meta-concept of faceted speech content, with text mining as a basic analysis method for linguistic data. By analyzing and comparing two types of data, interviews and community radio speech data, it was observed that the pattern of normative occurrence differed greatly depending on the interaction partner and that it was constant regardless of the interaction partner, indicating the necessity of an analytical frame based on a dynamic language perspective.

In planning the reallocation of road space such as shared spaces and Zone 30, it is essential to have a better understanding of the process of mutual norms based on the heterogeneity of human response. We proposed a model of behavioral norms generated in a square space by using an evolutionary game framework to investigate what kind of norms are generated by the discrepancies between pedestrians and vehicles due to asymmetries in speed and spatial cognition. We developed a game theoretic interdependent decision-making model by means of structured inference and conducted a design simulation to show that behavioral norms changed significantly before and after the renovation of the station square.

We focused on the rambling behavior in the Shibuya, and tried to identify the spatial indicators that affect rambling behavior on the network based on the precise observation results of human behavior in alleyways. This study gives meaning to the context of walking around the city, which has been ambiguously used in previous studies, by introducing the concepts of optional, necessary, and social into the rambling behavior, and by collecting data on human rambling in the Shibuya using the probe person technique. We estimated a sequential route choice model based on indexing the street potential of the Shibuya neighborhood, and clarified the influence of the design of a surplus space on rambling behavior.

We analyzed the process of modernization in Matsue by three different layers: the development of infrastructures in macro scale, the history of urban planning, and lastly the transformation of plots along the historic highway. The historic infrastructures in Matsue, water transportation, a historic highway, and channels, have been surviving until now because of the influence of disasters and the national law. By detailed analysis of the transformation process of plots along the historic highway, it is revealed that the speed and extent of transformation were different according to the accessibility to the transportation or the characteristics passed down from the era before modernization. By drawing the different process by organizing them in a typological way, we tried to illustrate the coexistence of historic urban landscape and modern convenience in Matsue.

In this research, in order to clarify the generation mechanism of the parking lot distribution in the city center, we formulated the parking lot location and the shop location at the same time and analyzed the equilibrium state. We showed that this equilibrium problem can be interpreted as a potential game by taking in the restriction of the parking lot that it is necessary to return to the place where it stopped, and specified a global stable state. As a result, in the agglomeration and dispersion of parking lot distribution in the city center, a bifurcation phenomenon may occur due to the two parameters, the property of the building located outside the parking lot and the size of the utility of the visitors' walking on foot.