Teleoperated driving complements automated driving and acts as transitional technology towards full automation. An economic advantage of teleoperated driving in logistics operations lies in managing fleets with fewer teleoperators compared to vehicles with in-vehicle drivers. This alleviates growing truck driver shortage problems in the logistics industry and save costs. However, a trade-off exists between the teleoperator-to-vehicle (TO/V) ratio and the service level of teleoperation. This study designs a simulation framework to explore this trade-off generating multiple performance indicators as proxies for teleoperation service level. By applying the framework, we identify factors influencing the trade-off and optimal TO/V ratios under different scenarios. Our case study on road freight tours in the Netherlands reveals that for any operational settings, a TO/V ratio below one can manage all freight truck tours without delay, while one represents the current situation. The minimum TO/V ratio for zero-delay operations is never above 0.6, implying a minimum of 40% teleoperation labor cost saving. For operations where a small delay is allowed, TO/V ratios as low as 0.4 are shown to be feasible, which indicates potential savings of up to 60%. This confirms great promise for a positive business case for the teleoperated driving as a service.

Automated driving developments should be considered when making decisions about investments in physical and digital infrastructure. This paper proposes four scenarios for automated driving developments in the Netherlands in 2040 and 2060 taking into account uncertainties regarding future penetration rates, the level of connectivity, the operational design domain, and the expected impacts of automated driving: 1) Late transition, 2) Automated vehicles on main roads, 3) Car-topia, and 4) Share-topia. To derive these scenarios, an extended switchboard method is introduced in which multiple driving forces for automated driving can be varied. The main driving forces were identified based on expert surveys. For each scenario, a modelling approach is used to compute the impact of automated driving on vehicle kilometres driven and congestion. The extended switchboard method offered more flexibility than existing scenario methods. The model-based impact assessment provided more conservative and probably more accurate insights into the expected impacts of automated driving on vehicle kilometres driven and congestion than expert estimates from the literature. The results show that in all scenarios automation leads to an increase in the number of trips, vehicle kilometres driven and congestion. In the scenarios with autonomous vehicles, congestion is expected to increase up to 17%. The higher the penetration rates of connected automated vehicles, the smaller the increase in congestion (1.5%-11%). The results indicate that investments in digital infrastructure are needed to prevent capacity reduction due to autonomous driving. The scenarios “car-topia” and “share-topia” may require additional physical infrastructure on motorways and regional roads, and/or the implementation of demand management strategies.

This study proposes a hybrid deep-learning-metaheuristic framework with a bi-level architecture for road network design problems (NDPs). We train a graph neural network (GNN) to approximate the solution of the user equilibrium (UE) traffic assignment problem and use inferences made by the trained model to calculate fitness function evaluations of a genetic algorithm (GA) to approximate solutions for NDPs. Using three test networks, two NDP variants and an exact solver as benchmark, we show that on average, our proposed framework can provide solutions within 1.5 % gap of the best results in less than 0.5 % of the time used by the exact solution procedure. Our framework can be utilized within an expert system for infrastructure planning to determine the best infrastructure planning and management decisions under different scenarios. Given the flexibility of the framework, it can easily be adapted to many other decision problems that can be modeled as bi-level problems on graphs. Moreover, we foreseen interesting future research directions, thus we also put forward a brief research agenda for this topic. The key observation from our research that can shape future research is that the fitness function evaluation time using the inferences made by the GNN model was in the order of milliseconds, which points to an opportunity and a need for novel heuristics that 1) can cope well with noisy fitness function values provided by deep learning models, and 2) can use the significantly enlarged efficiency of the evaluation step to explore the search space effectively (rather than efficiently). This opens a new avenue for a modern class of metaheuristics that are crafted for use with AI-powered predictors.

With the advent of automated vehicles (AVs), new infrastructure planning concepts such as subnetworks of AV-ready roads have been proposed to improve the performance of transportation networks and to promote the adoption of AVs. However, these subnetworks should evolve over time in response to the growing AV demand, which necessitates a multi-stage modeling approach. In this study, we propose multi-stage deployment of AV-ready subnetworks and formulate it as a time-dependent network design problem, which is a bi-level mixed-integer programming problem. The lower level is a simultaneous travel mode and route choice equilibrium with continuous decision variables, and the upper level is a design problem including infrastructure investment decisions to determine which roads to upgrade and include in AV-ready subnetworks for mixed traffic. We use a case study of a real road network to demonstrate the concept. Since computational efficiency is a key factor for solving such large-scale problems, we develop two efficient and tailored evolutionary heuristics to solve the problem, and compare their performance to a computationally demanding Genetic-algorithm-based solution method. The results indicate that the proposed algorithms can efficiently solve this large-scale problem while satisfying constraints in all scenarios, and outperform Genetic algorithm, particularly in the scenario with larger number of stages. Moreover, in all scenarios, deployment of AV-ready subnetworks leads to improvements in network performance in terms of total travel time and cost. However, the improvements are always accompanied with increased total travel distance. The extent of changes depends on AV market penetration rate, AV-ready subnetwork density and timing of densification.

Globalization and subsequent increase in seaborne trade have necessitated efficient planning and management of world's anchorage areas. These areas serve as a temporary stay area for commercial vessels for various reasons such as waiting for passage or port, fuel services, and bad weather conditions. The research question we consider in this study is how to place these vessels inside a polygon-shaped anchorage area in a dynamic fashion as they arrive and depart, which seems to be the first of its kind in the literature. We specifically take into account the objectives of (1) anchorage area utilization, (2) risk of vessel collisions, and (3) fuel consumption performance. These three objectives define our objective function in a weighted sum scheme. We present a spatio-temporal methodology for this multi-objective anchorage planning problem where we use Monte Carlo simulations to measure the effect of any particular combination of planning metrics (measured in real time for an incoming vessel) on the objective function (measured in steady state). We resort to the Simultaneous Perturbation Stochastic Approximation (SPSA) algorithm for identifying the linear combination of the planning metrics that optimizes the objective function. We present computational experiments on a major Istanbul Straight anchorage, which is one of the busiest in the world, as well as synthetic anchorages. Our results indicate that our methodology significantly outperforms comparable algorithms in the literature for daily anchorage planning. For the Istanbul Straight anchorage, for instance, reduction in risk was 42% whereas reduction in fuel costs was 45% when compared the best of the current state-of-the-art methods. Our methodology can be utilized within a planning expert system that intelligently places incoming vessels inside the anchorage so as to optimize multiple strategic goals. Given the flexibility of our approach in terms of the planning objectives, it can easily be adapted to more general variants of multi-objective spatio-temporal planning problems where certain objects need to be dynamically placed inside two or even-three dimensional spaces in an intelligent manner.