There are mainly two angles of view to investigate the traffic system, either from the view of traffic management, or from the view of individual travelers (or a group of homogenous travelers). The former, who designs and operates the system, is seeking for a solution with minimal total system cost, including vehicle en route travel time, emissions, delay, infrastructure construction and maintenance expenses, etc., and maximal facility utility, including safety, capacity, service level, etc. On the other hand, the latter, who rarely shares the same perspective with the former, is continually searching for and switching to the route with a better travel experience, for instance, with a shorter travel distance or travel time, lower travel cost, higher reliability, etc. Considering...[ Read more ]

There are mainly two angles of view to investigate the traffic system, either from the view of traffic management, or from the view of individual travelers (or a group of homogenous travelers). The former, who designs and operates the system, is seeking for a solution with minimal total system cost, including vehicle en route travel time, emissions, delay, infrastructure construction and maintenance expenses, etc., and maximal facility utility, including safety, capacity, service level, etc. On the other hand, the latter, who rarely shares the same perspective with the former, is continually searching for and switching to the route with a better travel experience, for instance, with a shorter travel distance or travel time, lower travel cost, higher reliability, etc. Considering these two different perspectives, this thesis develops adaptive traffic control and vehicle navigation methods, assuming the availability of real-time information, and that the traffic system is evolving from period to period as the traffic control and/or navigation system adapt and travelers modify their route choices dynamically.

Adaptive and dynamic are the two key words in this research. The term 'adaptive' carries two meanings: a), the ideas and objectives in this thesis are flexible and adaptive to different scenarios and objectives; b), the outputs of the formulations are relatively robust and cater for different real time information adaptively. By labeling the models and formulations as adaptive, we mean that they cater for different real time information and their sources such as travel time prediction from radios, posts from friends on social networks, trip suggestions from internet/web. Besides, we also mean that the decision making process is carried out adaptively while the trip is evolving. Similarly, the term 'dynamic' also counts in two aspects: a), the core technique used in the formulation is dynamic programming which grants the formulation the ability to adapt to different conditions; b), the subject studied in this research is viewed as a system evolving from period to period dynamically. By labeling our methods as dynamic, we imply that the dynamic programming techniques are used to assist in making trip/control decisions such that travelers have better trip experiences. At the same time, the traffic control measures and travelers in the traffic system are both considered dynamically evolving period to period.

This thesis is expected to make theoretical contributions to traffic control and vehicle navigation methods by extending the concepts of adaptive control and period-to-period dynamics for traffic management while taking advantage of real time traffic information. Furthermore, with a proper representation of traffic control measures and traffic flow assignment principle, coupled with an explicit description of the dynamical system evolution process, a more comprehensive picture of the traffic system performance can be formulated. It is hoped that such a deeper understanding in and linkage between travel behavior models and system control strategies will open up theoretically interesting, yet practically important areas for transportation system management.