This thesis concerns the mathematical modeling of transportation systems for improved decision support and analysis of transportation-related problems. The main purpose of this thesis is to develop and evaluate models and methods that exploit link flows. Link flows are straightforward to obtain by measurements or estimation methods and are commonly used to describe the traffic state. The models and methods used in this thesis apply mathematical optimization techniques, computer simulations, and probabilistic methods to gain insights into the transportation network under study and provide benefits for both traffic managers and road users.  First, we present an optimization model for allocating charging stations in a transportation network to serve owners of electric vehicles. The model utilizes a probabilistic route selection process to detect locations through which vehicles may pass. It also considers the limited driving range of electric vehicles. The iterative solution procedure finds the minimal number of minimal charging stations and their locations, which provides a lower bound of charging stations to cover each of the considered routes. Second, we present a case study, in which we argue that stationary and mobile measurement devices possess complementary characteristics. In that study, we investigate how speed cameras and probe vehicles can be used in conjunction with each other for the collection of detailed traffic data. The results show that the share of successfully observed and identified vehicles can be significantly improved by using both stationary and mobile measurement devices. Third, we present a simulation model with the intent of finding the most probable underlying routes based on hourly link flows. The model utilizes Dijkstra's algorithm to find the shortest paths and uses a straightforward statistical test procedure to find the most significant routes in the network based on replicated movements of trucks. Finally, we investigate the possibility to study how the traffic flow in one location reflects the flows in the surrounding area. The statistical basis of the proposed model is built upon measured link flows to study the dispersion of aggregate traffic flows in nodes. By considering the alternative ways vehicles can travel between locations, the model is able to determine the expected link flow that originates from a node in a nearby region. The results of the thesis show that the link flows, which are basic descriptors of the road segments in a transportation network, can be used to study a broad range of problems in transportation.

abstract: The autonomous vehicle technology has come a long way, but currently, there are no companies that are able to offer fully autonomous ride in any conditions, on any road without any human supervision. These systems should be extensively trained and validated to guarantee safe human transportation. Any small errors in the system functionality may lead to fatal accidents and may endanger human lives. Deep learning methods are widely used for environment perception and prediction of hazardous situations. These techniques require huge amount of training data with both normal and abnormal samples to enable the vehicle to avoid a dangerous situation. The goal of this thesis is to generate simulations from real-world tricky collision scenarios for training and testing autonomous vehicles. Dashcam crash videos from the internet can now be utilized to extract valuable collision data and recreate the crash scenarios in a simulator. The problem of extracting 3D vehicle trajectories from videos recorded by an unknown monocular camera source is solved using a modular approach. The framework is divided into two stages: (a) extracting meaningful adversarial trajectories from short crash videos, and (b) developing methods to automatically process and simulate the vehicle trajectories on a vehicle simulator. / Dissertation/Thesis / Video Demonstration / Masters Thesis Computer Science 2019

Collaborative dispatching allows several dispatchers to view the routing solution as a dynamic model where changes to the vehicle routes can be made in real-time. In this paper we discuss implications of collaborative dispatching on real-time decision support tools for motor carriers. We present a collaborative dispatching system which uses real-time information obtained from a telematics system. Messages sent from the vehicles are automatically analysed and actual data, such as exact arrival and departure times, as well as discrepancies between actual and planned data are identified. The collaborative dispatching system not only allows several dispatchers to concurrently modify the schedule, but also a dynamic optimisation method. The optimisation method is capable of taking into account that input data may change at any time and that dispatchers can concurrently modify the schedule and may add or relax certain constraints relevant to the optimisation model.

Real-life vehicle routing problems encounter a number of complexities that are not considered by the classical models found in the vehicle routing literature. In this paper we consider a dynamic real-life vehicle routing problem which is a combined load acceptance and generalised vehicle routing problem incorporating a diversity of practical complexities. Among those are time window restrictions, a heterogeneous vehicle fleet with different travel times, travel costs and capacity, multi-dimensional capacity constraints, order/vehicle compatibility constraints, orders with multiple pickup, delivery and service locations, different start and end locations for vehicles, route restrictions associated to orders and vehicles, and drivers’ working hours. We propose iterative improvement approaches based on Large Neighborhood Search. Our algorithms are characterised by very fast response times and thus, can be used within dynamic routing systems where input data can change at any time.

While the traditional approach to ease traffic congestion has focused on building infrastructure, the recent emergence of Connected and Automated Vehicles (CAVs) and urban mobility services (e.g., Autonomous Mobility-on-Demand (AMoD) systems) has opened a new set of alternatives for reducing travel times. This thesis seeks to exploit these advances to improve the operation and efficiency of Intelligent Transportation Systems using a network optimization perspective. It proposes novel methods to evaluate the prospective benefits of adopting socially optimal routing schemes, intermodal mobility, and contraflow lane reversals in transportation networks. This dissertation makes methodological and empirical contributions to the transportation domain. From a methodological standpoint, it devises a fast solver for the Traffic Assignment Problem with Side Constraints which supports arbitrary linear constraints on the flows. Instead of using standard column-generation methods, it introduces affine approximations of the travel latency function to reformulate the problem as a quadratic (or linear) programming problem. This framework is applied to two problems related to urban planning and mobility policy: social routing with rebalancing in intermodal mobility systems and planning lane reversals in transportation networks. Moreover, it proposes a novel method to jointly estimate the Origin-Destination demand and travel latency functions of the Traffic Assignment Problem. Finally, it develops a model to jointly optimize the pricing, rebalancing and fleet sizing decisions of a Mobility-on-Demand service. Empirically, it validates all the methods by testing them with real transportation topologies and real traffic data from Eastern Massachusetts and New York City showing the achievable benefits obtained when compared to benchmarks.

Intelligent Transportation Systems (ITS) can substantially improve roadsafety and trac eciency. This is possible by allowing communicationamong nearby vehicles and among vehicles and xed roadside units. A popularstandard for vehicular communications is IEEE 802.11p. It is basedon a CSMA/CA MAC method that does not guarantee channel access in anite time and so is not suitable for real-time communications. It also needsmethods to control and limit the load, since the transmission of periodicinformation among vehicles can saturate the channel. In this thesis, a newreal-time scheduling algorithm suitable for ITS applications is introduced. Itis based on a TDMA MAC method, where the roadside unit has the tasks toestimate the channel conditions and assign fractions of time slot to users. Alinear programming approach is considered to minimize an index of utility ofthe transmissions. Multi-hop communication scenarios among the vehiclesare considered for both uplink and downlink communications. It is shownhow the optimal duration of the fraction of time slot depends on the channelconditions. A higher channel gain corresponds to a higher transmission timewhereas a lower channel gain corresponds to a lower transmission time. Itis concluded that the approach studied in the thesis can guarantee a highutility provided that the complexity of the optimization is reduced as thenumber of involved vehicles increases.

This thesis presents the use of microscopic traffic simulation models to evaluate the effects of operational-level transportation projects such as ITS. A detailed framework outlining the construction and calibration of microscopic simulation models is provided, as well as the considerations that must be made when analyzing the outputs from these models. Two case studies are used to reinforce the concepts presented. In addition, these case studies give valuable insight for using the outlined approach under real-world conditions. The study indicates a promising future for the use of microsimulation models for the purpose of evaluating operational-level projects, as the theoretical framework of the models is sound, and the computational strategies used are feasible. There are, however, instances where simulation models do not presently model certain phenomena, or where simulation models are too computationally intensive. Comprehensive models that integrate microscopic simulation with land use planning and realistic predictions of human behavior, for instance, cannot practically be modeled in contemporary simulation packages. Other than these instances, the largest obstacles to using simulation packages were found to be the manpower required and the complexity of constructing a model. Continuing research efforts and increasing computer speeds are expected to resolve the former issues. Both of the latter concerns are alleviated by the approach presented herein. Within the approach framework detailed in this thesis, particular emphasis is given to the calibration aspects of constructing a microscopic simulation model. Like the simulation process as a whole, calibration is both an art and a science, and relies on sound engineering judgement rather than indiscriminate, formulaic processes. / Master of Science

Optimal control of autonomous systems is a fundamental and challenging problem, especially when many stringent safety constraints and tight control limitations are involved such that solutions are hard to determine. It has been shown that optimizing quadratic costs while stabilizing affine control systems to desired (sets of) states subject to state and control constraints can be reduced to a sequence of Quadratic Programs (QPs) by using Control Barrier Functions (CBFs) and Control Lyapunov Functions (CLFs). Although computationally efficient, this method is limited by several factors which are addressed in this dissertation. The first contribution of this dissertation is to extend CBFs to high order CBFs (HOCBFs) that can accommodate arbitrary relative degree systems and constraints. The satisfaction of Lyapunov-like conditions in the HOCBF method implies the forward invariance of the intersection of a sequence of sets, which can then guarantee the satisfaction of the original safety constraint. Second, under tight control bounds, this dissertation proposes an analytical method to find sufficient conditions that guarantee the QP feasibility. The sufficient conditions are captured by a single state constraint that is enforced by a CBF and then added to the QP. Third, for complex safety constraints and systems in which it is hard to find sufficient conditions for feasibility, machine learning techniques are employed to learn the definitions of HOCBFs or feasibility constraints. Fourth, when time-varying control bounds and noisy dynamics are involved, adaptive CBFs (AdaCBFs) are proposed, which can guarantee the feasibility of the QPs if the original optimization problem itself is feasible. Finally, for systems with unknown dynamics, adaptive affine control dynamics are proposed to approximate the real unmodelled system dynamics which are updated based on the error states obtained by real-time sensor measurements. A set of events required to trigger a solution of the QP in order to guarantee safety is defined, and a condition that guarantees the satisfaction of the HOCBF constraint between events is derived. In order to address the myopic nature of the CBF method, a real-time control framework that combines optimal trajectories and the computationally efficient HOCBF method providing safety guarantees is also proposed. The HOCBFs and CLFs are used to account for constraints with arbitrary relative degrees and to track the optimal state, respectively. Eventually, an optimal control problem based on the proposed framework is always reduced to a sequence of QPs regardless of the formulation of the original cost function. Another contribution of the dissertation is to apply the above proposed methods to solve complex safety-critical optimal control problems, such as those arising in rule-based autonomous driving and optimal traffic merging control for Connected and Automated Vehicles (CAVs).

No description available.

In this dissertation, we present generalized techniques which allow for the tracking and classification of vehicles by tracking various Point(s) of Interest (PoI) on a vehicle. Tracking the various PoI allows for the composition of those points into 3D geometries which are unique to a given vehicle type. We demonstrate this technique using passive, simulated image based sensor measurements and three separate inertial track formulations. We demonstrate the capability to classify the 3D geometries in multiple transform domains (PCA & LDA) using Minimum Euclidean Distance, Maximum Likelihood and Artificial Neural Networks. Additionally, we demonstrate the ability to fuse separate classifiers from multiple domains via Bayesian Networks to achieve ensemble classification.