The activity-based approach to travel demand analysis recognizes that human activities dictate travel. Microsimulation of household activity patterns has gained significant attention as a method for modeling this activity participation. Existing approaches, however, focus on modeling how households solve the activity scheduling problem—how and when each household member should engage in particular activities to meet the needs of the household. This is a top-down approach that recognizes inherent causal links between members of a household but sacrifices the modeling flexibility needed for complex policy analysis.
This dissertation synthesizes dominant activity analysis theories with concepts from the social simulation and complex systems analysis literature to demonstrate that the motivation and constraints that shape activities are more directly embodied in the activity execution problem—how individuals interact with other entities in their environment to engage in activity. The scheduling problem is re-cast as the adaptive internal process that an individual uses to navigate through this interactive environment to achieve environmentally-derived payoffs.
Based on this theory, I describe a microsimulation approach that focuses on the activity execution process. This bottom-up approach presents a problem of tractability. I solve this problem by describing activity execution using a model of negotiated interaction derived from the Contract Net Protocol for distributed computation. This model is shown to be tractable in terms of the number of negotiating individuals, given reasonable limitations on the negotiation process. Then, I describe a complete agent-based model of an urban activity system based on this activity execution kernel. This general model is shown to be tractable in terms of the population size, given assumptions on how negotiations are initiated. Finally, I present results from experiments using candidate adaptive learning algorithms for agents operating in the microsimulation to demonstrate the utility of the approach.
This dissertation is an initial feasibility study of a self-organized, distributed traffic information system, called “Autonet,” that is based upon peer-to-peer information exchange among vehicles with inter-vehicle communication equipment. Autonet, a concept proposed within the Cal(IT)2 Transportation Layer of the University of California, Irvine, is defined as an autonomous, self-organizing information network and control system for effective management of interactions among intelligently-informed vehicles, roadways, and stations. Before the proposed Autonet system can be implemented in a real-world transportation system, hardware and software requirements need to be identified; ideally, based on predictions provided by mathematical formulations and testing with microscopic traffic simulations. The research in this dissertation focuses on the traffic aspects of the proposed Autonet, using simulation approaches to assess the potential benefits that might be accrued by the traffic system and to evaluate the ability of the computing overlay to handle the traffic “application”.
An existing microscopic traffic simulator, which is treated only as the vehicle mover, is selected and integrated with originally developed inter-vehicle communication modules through application programing interfaces to build the simulation framework for the feasibility analysis of the proposed Autonet system. Traffic-related information propagation in the traffic network via inter-vehicle communication, which is the foundation for the proposed self-organized, distributed traffic information system, can be tested in detailed modeling under that simulation framework. This dissertation investigates traffic information propagation both in one-dimensional highway/freeway networks including one-direction and two-direction cases, and in two-dimensional arterial street networks, considering various roadway formats and incident conditions, for different combinations of modeling parameters related to the the proposed systems.
Further, a series of vehicle re-routing applications under the incident condition based upon the proposed self-organized, distributed information system are tested via simulation. Analysis of the simulation results is given for individual groups of vehicles and for the whole system to find potential benefits from Autonet implementation. Finally, this dissertation identifies needs for future research both for the modeling effort and for some issues involving actual implementation.
This dissertation focuses on the bidding problem in combinatorial auctions with particular application to the freight transportation contract procurement problem. Combinatorial auctions are auctions in which a set of heterogeneous items are sold simultaneously and in which bidders can bid for combinations of items. These auctions involve many difficult problems for both auctioneers and bidders and have received recent attention from computer scientists, operations researchers and economists. Large shippers have begun to use this method to procure services from trucking companies by selling contracts for delivery routes.
This dissertation analyzes the impact of the combinatorial auction procurement method on both shippers and carriers. I demonstrate that bidders have more complicated optimization problems than auctioneers in combinatorial auctions. The bid construction problem, i.e. how bidders identify and construct beneficial bids, is very difficult to solve and remains an open question. This dissertation investigates this problem and proposes an optimization-based approximation method that involves solving an NP-hard problem a single time, yielding significant improvements in computational efficiency.
I also examine the current state of the trucking and third party logistics industries. The trucking industry is very competitive and small carriers are operating under thin margins. This dissertation addresses this issue by proposing an auction-based collaborative carrier network in which participating carriers identify inefficient lanes and exchange them with partners. This is shown to be Pareto efficient. I then discuss decision problems related to how carriers identify inefficient operations and make and select bids. This represents an effort to utilize an advanced auction mechanism to enhance carriers’ operational efficiencies in e-commerce.
This dissertation research proposes the development and evaluation of a new concept for high-coverage point-to-point transit systems (HCPPT). The system is a radically new operational scheme for mass transit that relies on a large number of small transit vehicles routed using advanced communication technology and real-time stochastic control algorithms. The defense presentation covers the three major contributions of this research. First, the details of the concept are described, outlining its features and the flexibility of potential operational schemes. Second, the real-time routing algorithms are presented. A strict optimization formulation and solution for such a problem is computationally prohibitive in real-time. The design proposed in this dissertation is effectively geared towards a decomposed solution using detailed rules for achieving vehicle selection and route planning. If real-time update of probabilities based upon modeling the future dispatch decisions is included, then this scheme can be considered as a form of quasi-optimal predictive-adaptive control problem. Finally, some numerical results obtained from a simulated case study in Orange County are shown. This study was carried out using a multi-purpose simulation platform developed in the context of this research. The final simulations required point-to-point vehicle simulation, which is not possible with off-the-shelf simulators. The simulation framework uses a well-known microscopic traffic simulator that was significantly modified for demand-responsive vehicle movements and passenger tracking. A brief demonstration (animation) of the simulation tool applied to the HCPPT scheme is also included in the presentation.
Most economists agree that new investments in highways at this point in time in the United States have little impact on overall growth in output. New highways play a more important role in shifting economic activities among places, drawing jobs from other locations into the highway corridors, a phenomenon known as negative spillovers. The objective of this dissertation is two-fold, to examine the proposal to decentralize highway finance, which aims to solve the financial responsibility mismatch problem that stems from economic spillovers of highways, and to test the hypothesis of economic spillovers of highway investment at the metropolitan level. First, to better understand how spillovers influence the highway investment decision, the theoretical framework from the interjurisdictional tax competition literature is borrowed to model governments’ investment behaviors. Numerical simulations show that decentralized local governments, which independently maximize output in their own jurisdiction, may engage in wasteful investments in highway with the presence of spillovers. Second, to shed more light on the spatial detail of economic spillovers, empirical tests of the spillovers hypothesis are conducted at the metropolitan level, with census tracts as the unit of observation. The results of the quasi-experiment reveal the employment growth patterns of census tracts in Los Angeles County that confirms the existence of negative spillovers, caused by the opening of the Interstate 105 in 1995. The benefiting area, which grew substantially after the highway was opened, is limited to a long narrow highway corridor while nearby locations outside the corridor experience slow growth relative to the rest of the metropolitan area after controlling for various factors. Together, these results suggest that although negative spillovers are present at the metropolitan level, decentralizing highway finance may not be an effective policy to deal with the financial responsibility mismatch problem. Highway finance should remain centralized within metropolitan areas, and the regional governing bodies should pay special attention to the distributional impact of highway projects.
One of the main barriers to a better understanding of activities and
travel patterns is the difficulty in collecting long-duration data.
Previous studies have examined computer-aided interview techniques.
Others have researched the potential for global positioning system
(GPS) antennas to collect more accurate travel data. This
dissertation adds the use of wireless communications technology to
integrate GPS data with a dynamically generated, web-based activity
survey. In addition, three separate analysis techniques are examined
using the results of an informal pilot test. The goal of these
analysis techniques is to weave together the large set of GPS data
that can be collected with the much smaller set of activity
responses. The net result represents both an advance in data
collection techniques, as well as a new, peer-to-peer approach to
gathering and sharing experiential transportation information, an
approach that should be incorporated into future Intelligent
Transportation Systems designs.
Use of Advanced Traveler Information Systems (ATIS) are considered a
promising way to improve traffic condition by helping travelers to
efficiently use existing transportation facilities. The research
examines a wide variety of information dissemination schemes under
technologies such as in-vehicle navigation systems, changeable message
signs, GPS-based location systems and wireless or Internet based
vehicle communication and routing. This study evaluates various route
guidance systems via static and dynamic network optimization and
traffic simulation models. Parametric studies are conducted on certain
aspects, due to the lack of good models on driver response/compliance
to ATIS information. The dissertation also develops preliminary
insights on networks with multiple information service vendors and the
complex dynamics that result from it. The research methodology
incorporates non-linear network optimization algorithms, heuristic
optimizations as well as traffic network simulation schemes.
The process of activity scheduling is crucial to the
understanding of travel behavior changes. In-depth research is urgently
needed to unearth this process. To reveal this process, a new computer
program, REACT!, has been developed to collect household activity
scheduling data. The program is implemented as a stand-alone program with
Internet connectivity for remote data transmission. It also contains a GIS
for location identification and a special feature that traces the
decisions in scheduling process. A pilot study was conducted in Irvine,
California to evaluate the program performance. Experience from the pilot
study validated the program’s capability of guiding participants to
complete data entry tasks on their own, thus the objective of reducing the
cost and human resource of such a computerized survey is achieved. Other
positive results regarding objectives of reducing instrumental biases and
expanding program capabilities were also obtained. Areas for improvement
were also identified.
Based on the pilot data, activities with shorter duration were found more
likely to be opportunistically filled in a schedule already anchored by
their longer duration counterparts. In addition, the situations (e.g.,
location, involved person, and day of the week) under which an activity
occurred were found related to its scheduling horizon. Analyses were also
performed to validate that the above findings hold in the presence of a
third factor (i.e., in-home vs. out-of-home, and work/school vs.
non-work/school). Additionally, analysis of tour structure reveals that a
certain portion of trip-chains was formed opportunistically. The
proportion of opportunistic stops tends to increase as stop sequence
increase. Travel time required to reach an activity is also positively
related to scheduling horizon of the activity, with distant stop being
planned earlier.