Browsing by Author "Roncoli, Claudio, Prof., Aalto University, Department of Built Environment, Finland"

Now showing 1 - 2 of 2

Expanding the applicability of large-scale transportation models for the assessment of disruptive mobility technologies
(2024) Agriesti, Serio Angelo Maria
School of Engineering | Doctoral dissertation (article-based)
As the world grows more complex and most aspects of daily life become fluid and more subject to change, transportation is no exception. Urbanization, sprawling, inequalities, and climate change are only a few of the challenges currently facing transportation planners, institutions, and public bodies. Moreover, historical patterns become less reliable as disruptions that once were counted in decades are now happening every few years. To address all these challenges, it is of utmost importance to adapt our approaches to be more flexible and to frame changes in attitudes, utilities, and goals in the urban population.The work presented in this thesis focuses on developing and providing multiple solutions with a specific focus on large-scale urban models. The main objective is to draw a road map of the major issues hindering modeling solutions able to tackle the challenges described above. Once that is accomplished, different methods are designed, developed, and tested on a real case study (Tallinn, the capital city of Estonia). First, we approach the problem of defining a synthetic population detailed enough to carry out large-scale behavioral studies, without infringing privacy constraints. We then use the resulting dataset to build an activity-based behavioral model for the whole city. We harness machine learning techniques to automate the calibration of the hundreds of behavioral parameters involved, a quantity not yet achieved in the state of the art. We then focus on integrating the behavioral model with state-of-the-art traffic assignment solutions, trying to forge a blueprint for any modeler wanting to expand an existing model (a problem quite common, as many urban traffic assignment models have been developed through the years and are not easily replaceable). An iterative approach is developed and tested, first to frame a baseline situation and then to forecast the impact of disruptive mobility services on both the demand and the supply. Finally, the large-scale urban architecture built by integrating behavioral and traffic assignment models is exploited to test, for the first time with the presented degree of detail, the impacts of both an optimization algorithm and a fairness pricing scheme on an (automated) on-demand system.
Optimal and adaptive controller design for motorway traffic with connected and automated vehicles
(2023) Tajdari, Farzam
School of Engineering | Doctoral dissertation (article-based)
Connected vehicles control has brought the opportunity of using a higher level of Traffic Management Center (TMC) to investigate the driving-related terms of congestion avoidance, and safety and pleasantness of driving. Apparently, the efficiency of the terms is improved by the true identification of the assumptions made by the TMC. This thesis adopts microscopic and macroscopic investigations in a motorway considering the effects of lane-changing and ramp metering that exploits the presence of connected vehicles. Microscopically, a car-following behavior in practical scenarios may witness some special situations affected by the vehicles in the adjacent lane that diverts the behavior from the conventional car-following conditions, e.g., a Follower Vehicle (FV) behavior witnesses a lane changing.Addressing the complex situation, human drivers practically may anticipate the surrounding vehicles before lane-changing to make safe control decisions for the behavior. Thus,we design a novel human-like fuzzy controller for the FV witnessing the exiting lane-changer in the complex behavior subject to comfort and safety. The approach will improve the conventional car following Advance Driving Assistant Systems (ADAS), towards a fully automated car-following that enhances the connected vehicles' frameworks. From a macroscopic perspective, few studies may actually have a direct impact on traffic flow control among the wide range of available systems, while the majority of them aim at primarily improving safety or driver convenience. To address these issues, we bridge conventional traffic control, i.e., ramp-metering, with lane-changing control enabled by vehicle automation via a novel feedback-based integrated control strategy. We envision that, if lane-changing control capabilities are implemented in conjunction with more traditional traffic management strategies, such asramp-metering, in an integrated fashion, the resulting effectiveness, in terms of traffic performance, would be further increased. This allows us for rigorous investigations on the generalisability and robustness of the methodology for different network typologies and under different disturbances. However, the integrated lane-changing and ramp-metering control approach assume predefined constant set-points, typically critical density, which is supposed tobe determined based on, e.g., historical data. On the contrary, even if a set-point is known, it may not always be optimal due to possible changes in traffic behavior characteristics, such as, e.g., adifferent traffic composition. Thus to complete the control loop for general usage, we need to designan adaptive estimator integrable to the control strategy capable to update accurate critical density (constant or time-varying).