The widespread adoption of Industry 4.0 technologies is resulting in a wide availability of real-time data gathered on the shop floor. This data, once properly elaborated, can be used to support dynamic decision-making, im-proving manufacturing companies’ capability to deal with uncertainty and thus leading to potential benefits in their performance. This paper presents a simulation model to assess the changes in manufacturing systems perfor-mance resulting from the use of real-time data in the dynamic scheduling of in-plant logistics activities. The model was developed considering a general factory layout and implemented in Python, a widely used open-source pro-gramming language. Therefore, the model can be used and extended by a wide community of researchers, serving as a base for future studies, and adapted to be applied to a large number of factories, thus favoring a more widespread adoption of dynamic scheduling systems in practice. In this study, the model was applied to the setting of a factory in the food industry in which a fleet of mobile robots supply materials to production stations and retrieve finished goods, carrying them to the factory warehouse. Results show that a dynamic scheduling system, in which in-plant logistics activities are scheduled considering real-time data on the status of shop floor re-sources, leads to better performance, in terms of production stations uptime, compared with the static system currently adopted by the company.

Automated Guided Vehicles (AGVs) and Autonomous Mobile Robots (AMRs) are flexible and reliable options for material handling automation. The integration level with the roduction/logistic systems is crucial for performance and investment costs. Proper design of loading/unloading points is essential as they impact the number, level of automation, sorting/buffering level, and vehicle requirements. This paper presents an innovative approach combining virtual-interactive simulation and mathematical modeling to optimize loading/unloading points for maximum operational and economic performance. This approach simulates different scenarios and identifies the best loading/unloading points configurations optimizing the whole system’s performance. A numerical analysis is reported to demonstrate the practical implications.

The integration of Autonomous Guided Vehicles (AGVs) into smart facto-ries is transforming modern manufacturing, creating coexistence between humans and robotic systems. In this evolving landscape, one critical aspect is the seamless coordination of AGVs and human workers within factory set-tings. To achieve this, our research presents a portable indoor localization system that utilizes ESP32 microcontrollers as compact access points. Using Wi-Fi Fine Time Measurement (FTM) with smartphones, the system esti-mates worker positions through multi-lateration techniques in conjunction with advanced filtering methods. This localization system serves as a pivotal bridge, ensuring that AGVs can interact with and respond to the movements of shop floor workers. A field study in an actual warehouse environment val-idates the system's performance, demonstrating 1.13 meters accuracy in lat-eral movements. Furthermore, its localization capabilities within specific warehouse areas showcase its potential to enhance order picking processes and optimize human-AGV interaction.

The dynamic and rapidly evolving business environment poses numerous complexities for production logistics. The increasing frequency of product and model changes, coupled with the growing variability of components, underscores the urgency for adaptive measures to address ever-shortening product life cycles and advancing customer demands. Technological advancements have enhanced logistical productivity, but it is important to comprehensively tackle these challenges. To overcome these limitations, the concept of "transformability" is explored as a central cornerstone of the solution.

A design framework is proposed to increase the transformability. This paper systematically captures and models the production logistics system to achieve the goal. Key change enablers are identified and aligned with the production logistics system to develop specific change enablers for each production logistics area. These serve as the foundation for formulating transformation measures that can be incorporated into logistics planning. The results guide companies for successful implementation and long-term competitiveness by offering potential users a wide range of possibilities to increase their transformability within planning activities. This work contributes to raising awareness of the importance of transformable production logistics and offers practical recommendations for action. By embracing this approach, companies can proactively and effectively respond to dynamic fluctuations in the production environment, ensuring long-term competitiveness and sustainability

This study set out to develop a Reinforcement Learning (RL) agent for solving an extended Permutation Flow Shop Scheduling Problem (PFSSP). From the domain perspective, we see a lack of realistic constraints for synchronized, human-centered assembly lines. Moreover, objective functions must be provided to enable stress-reducing as well as robust planning under uncertainty. From a methodical perspective, RL has received more and more attention for problems of this type. However, we cannot identify applicable RL concepts for our extended PFSSP with multicriteria objectives. We propose a generic RL agent, which operates on an abstract representation of the schedule and with an objective-independent reward function. Our numerical experiments demonstrate that the agent successfully generalizes a policy and achieves better scores than a Simulated Annealing (SA) metaheuristic.

Manual work is a significant cost driver in manufacturing and logistics. However, research on the methods for analyzing manual processes utilizing sensor technologies, apart from technical feasibility, is scarce. Motion-Mining® is a technology that uses motion sensors, Bluetooth, and pattern recognition to enable highly automated process mapping and analysis of manual work.

The aim of this paper is to evaluate the benefits and limitations of applying this technology in manual production and logistics processes. To this end, Motion-Mining® is compared with traditional and low-tech Lean management tools for capturing and analyzing manual activities.

Ten semi-structured expert interviews as well as case studies in four companies were conducted. The results indicate that Motion-Mining® differs from Lean tools for process analysis mainly in terms of the effort required for data collection, the amount of data obtained, the representativeness of the data, the level of detail, and the insights gained.

This real-world simulation study analyzes a newly planned factory layout of a production company. Therefore, the first goal is the validation of the layout concerning bottlenecks, e.g., buffers, machines, and the planned trans-portation organization. The second goal is the analysis of different possible im-provements, regarding scheduling (different priority rules) and automatic trans-portation with AGVs. The best number of AGVs in terms of cost and logistic service level for the selected scenario is determined by the simulation study. Scheduling methods for jobs and AGVs are also compared, since they have high impact on the goal criteria, e.g., lead times. The study shows, that the se-lected layout including machine capacities is able to handle the estimated amount of occurring jobs in the future. Furter, an effective setup for the scenar-io could be found, which also supports the requirements of flexibility.

Steelmaking through Electric Arc Furnace (EAF) is known to be energy and cost intensive therefore, any improvement in processes will result in eco-nomic and environmental benefits. This study aims to improve the efficien-cy of the EAF process by predicting the most feasible steel grade which can be obtained with minimum purification based on endpoint parameters. Na-ïve Bayes classifier algorithm was employed to categorize EAF operational data. The operational data consists of 16 parameters with more than ten thousand data samples which classified into 6 possible steel grades, the carbon content of molten steel is defined as the decision variable to classify operational data. Finally, the results are also compared with MS Excel to examine how well the machine learning algorithm can be obtained. The re-sults show the algorithm can classify data with more than 90% accuracy.

In robotic mobile fulfillment systems, which are warehousing technologies that follow the parts-to-picker concept, the order picking process involves two decisions of how to schedule the processing of orders and of how to sequence the racks that are lifted and transported by robots to the picking station in order to supply the requested items. We propose a heuristic solution approach for solving the order-scheduling and rack-sequencing problem at a single picking station. Our approach utilizes column generation to partition the set of orders into batches. The goal is to minimize the number of rack assignments to these batches, which minimizes the rack-visits. The generated batches possess a specific property that allows for the straightforward derivation of an order-processing schedule and rack sequence. To further improve the solution, we refine the heuristic approach by rearranging the processing of batches and their assigned racks. We conducted a comprehensive and comparative computational. The method outperforms several other heuristics in terms of both solution quality and runtime on the majority of instances. Additionally, our heuristic yields satisfactory results when embedded into a framework designed to solve the problem across multiple picking stations, particularly for small-case data.

In order to provide fast access times to the stored items in warehouses, different types of storage systems exist. A number of factors, including the physical size and weight of the items to be stored, the frequency of use, and the resources (such as space) available, will determine the type of system that is suitable for a particular warehouse. In this talk, we will examine a new type of storage system that has received little attention from the scientific community so far, the compact storage systems. Compact storage systems aim at achieving the highest possible space utilization rate given limited storage space.

We want to shed light on the specifics of such storage systems and analyze how to optimize the storage and retrieval of items within them (e.g., by order picking and batching, the sequencing of the retrievals, and control, navigation, and relocation of storage units). In this talk, we formulate combinatorial optimization problems addressing the retrieval of items, discuss algorithmic approaches and their computational complexity.

Delivery of parcels in urban environments can be conducted in various ways. Next to traditional van delivery, cargo bikes are in use already today. In the future, also robots or drones may be used for this purpose. Each of those systems comes with certain (dis-)advantages regarding capacity, speed, environmental friendliness, and compatibility with customer preferences. The availability of these alternative transportation systems therefore raises the question, which of them to apply in a particular city environment and how to combine complementary systems in order to benefit from their mutual advantages. In order to investigate this, we propose a mathematical optimization model for the operations management of two-tier urban parcel logistics, which can handle alternative modes of transportation in isolation but also in combination. We propose an Adaptive Large Neighborhood Search for solving this optimization model and present computational results that shed light on the performance of the heuristic as well as the suitability of combined fleets of vans, bikes, robots and/or drones in urban environments. Through this, we can draw recommendations on which technologies to consider when implementing innovative parcel delivery systems for urban environments.

Transport times in container terminals have been widely studied, often with a focus on autonomous vehicles, while manual operations remain prevalent in many ports. Predicting travel times for straddle carriers in container terminals is a challenging problem, impacting the efficiency and productivity of manual container handling.

In our work, we propose a machine learning-based method for predicting travel times, leveraging a unique dataset obtained from a digital twin. This dataset incorporates positional and operational data gathered from IoT devices. The dataset is sourced from the Port of Hamburg, which is one of the 20 largest container ports in the world.

Our method demonstrates better accuracy and performance compared to conventional methods, tailored to the needs of manual container handling. The conducted research provides insights for optimizing container handling in container terminals. It also has implications for the container terminal simulation and an underlying digital twin.

The Capacitated Vehicle Routing Problem with Time Windows (CVRPTW) is a well-known combinatorial optimization problem that extends the classical Vehicle Routing Problem to account for additional real-world constraints such as truck capacity and customer time windows. Recent studies have explored the use of deep graph convolutional networks (GCNs) to predict the arcs that are part of the optimal tour for the Travelling Salesman Problem and related routing problems.

In our talk, we propose a novel context complemented graph convolutional network (CCGCN) which is integrated into a Monte Carlo Tree Search (MCTS) to solve the CVRPTW sequentially. The CCGCN consists of a deep convolutional part that builds efficient CVRPTW graph representations and a context part. The context part processes information of partially built tours to output probabilities on which vertex to add next to the tour, which is used during the expansion of the search tree. For simulating the final solution value based on a given partial solution in the Monte Carlo search tree, we apply a beam search that uses the context complemented graph convolutional network in an autoregressive form to build valid complete solutions for evaluating a given node in the MCTS. Moreover, a variation of the Upper Confidence Bound applied to trees is used in combination with the prediction probabilities given by the network to manage the tradeoff between exploration and exploitation. Evaluations of the proposed heuristic are performed on benchmark instances of the CVRPTW with up to 100 customers; these show promising results.

Online tour planning in logistics involves dynamically incorporating new transportation requests into precomputed tour plans, without foresight into future demands. Prominent examples include medical logistics involving the timely delivery of medical equipment, pharmaceuticals, or even patients. Online decision-making is quite well understood from an algorithmic and worst-case perspective and tight performance bounds are known.

However, the assumption of not having any prior knowledge about future requests seems overly pessimistic. Given the success of machine-learning methods and the available data in many tour planning problems, one may expect to have access to predictions about future requests. However, simply trusting them might lead to very poor solutions as these predictions come with no quality guarantee. In this talk we present recent developments in the young line of research that integrates such error-prone predictions into algorithm design to break through worst case barriers. We discuss algorithmic challenges with a focus on online routing present algorithms with error-dependent performance guarantees and we shortly discuss the choice of error metrics.

This is an overview talk that builds mainly on joint work with G. Bernardini, A. Marchetti-Spaccamela, L. Stougie, and M. Sweering, published at NeurIPS 2022.

Global trade heavily relies on maritime transport, with ships carrying most of the world's goods by volume. The effective planning of ports plays a pivotal role in the smooth operation of intermodal hubs required for the timely flow of goods. Ports are dynamic entities, constantly evolving due to technological advancements, economic fluctuations, political changes, and environmental factors, demanding innovative approaches to port planning. Furthermore, the complexity of port planning is compounded by the lengthy concessions that typically span 20 to 30 years.

The project HafenPlanZEN aims at providing a tool to facilitate and improve port planning decisions. The initiative represents a collaborative endeavor involving the Hamburg Port Authority, Hamburg Port Consulting, and the University of Hamburg. HafenPlanZEN harnesses this digital infrastructure, integrating data from various digital twins, offering a comprehensive overview of the port's performance and efficiency.

Central to our project is the development of a simulation and optimization tool aimed at the assessment and enhancement of port performance. HafenPlanZEN adopts a simulation-based approach and is directly fed with data from the port’s sensors and digital twins. Simulations allow port planners to experiment with diverse new development ideas as well as their refinement using optimization techniques. Such techniques can be used to automatically aid with infrastructure-independent decisions such as timing traffic lights and defining minimal parking and handling areas. Moreover, HafenPlanZen's capabilities extend to evaluating the impact of infrastructure changes, such as the construction of a new bridge or tunnel, providing comprehensive insights into port planning and management.

The aim of berth allocation planning is to derive conflict-free vessel assignments to the quay of a container terminal. An important objective of terminal operators in this context is to provide the best possible service quality to the shipping companies, i.e., especially short waiting times. The berthing schedule resulting from solving a dynamic Berth Allocation Problem (BAP) consists of the berthing times and positions of all vessels that are expected to arrive within a certain timeframe; these vessels are scheduled according to their respective arrival and handling times. However, both these times are uncertain due to different influences, e.g., wind and wave or defect handling equipment. Deviations from the planned handling time lead to delayed vessel departures, which cause waiting times for the succeeding vessels and also can ultimately result in conflicts that may impede the schedule’s feasibility. Hence, updating or re-planning of berthing schedules can become necessary, but this is costly and may be impossible when a plan is already in execution.

Therefore, the aim of this work is to derive robust berthing schedules that enhance the schedules’ stability by considering uncertainty already in the planning phase and, thus, are resistant to uncertainties of handling times. With a robust optimization approach which is based on time buffers, uncertainty is proactively considered, resulting in more robust schedules. The results of the new approach are evaluated from an ex post perspective using real ship data from the AIS and actual ship handling times.

Logistics operations often involve activities of different vehicle classes such as trucks, cars, and scooters. Due to their heterogenous physical properties, the different vehicle classes vary in their impact on the congestion of road networks. In this talk, we study the game-theoretic framework of atomic congestion games with different user classes. In these games, users control traffic composed of different vehicle classes and strive to minimize their travel time in the network. We discuss the existence of pure Nash equilibria in these games. To this end, a set of cost functions is called consistent for this class if all games with cost functions from the set have a pure Nash equilibrium. We give a complete characterization of consistent sets of cost functions showing that the only consistent sets of cost functions are sets of certain affine functions and sets of certain exponential functions. This characterization gives an axiomatic justification of the passager-car-unit concept used frequently in the traffic literature and can be extended to a larger class of games where each atomic player may control flow that belongs to different classes.

A very important task in modern logistics is the estimation of the arrival time of a planned transport. To give precise answers there is a huge demand for accurate traffic prediction models, especially for truck transportation that is performed on public roads. As such, there has been a huge effort to understand congestion both using theoretical models as well as simulations. This work focuses on the theoretical part. For theoretical traffic models, there is strong motivation to push them to be as realistic as possible.

The theoretical traffic model with dynamic time that gained the most attention in recent years is the deterministic fluid queuing model, already introduced by Vickrey.

In this work, we extend the deterministic fluid queuing model with a multi-criteria objective function. We assume that users try to minimize costs subject to arriving at the destination before a given deadline. Here, costs could be thought of as an intrinsic preference a user has regarding the different route choices, and queuing dynamics only play a role in the arrival time of a user.

We determine the existence and the structure of Nash flows over time and fully characterize the price of anarchy for this model, which measures the ratio of the quality of the Nash flow and the optimal flow.

Customer withdrawals from contracts can impact operational efficiency and resource utilization across various logistics scenarios. In the shipping and freight industry, for instance, customers may withdraw from shipping contracts due to shifts in their business needs, changes in market conditions, or unforeseen circumstances. This often necessitates adjustments in cargo consolidation and vehicle assignment. Similarly, in the last-mile delivery for e-commerce, customers may cancel orders or alter delivery preferences, leading to the need for reoptimized delivery routes and adjustments in resource allocation, such as bookings for the vehicle fleet, to ensure efficiency and cost-effectiveness. While individual contract cancellations may create room for new customers (orders, bookings), the paramount concern is the fulfillment of contracts with the remaining customers.

We address the problem of reoptimization with a commitment requirement by introducing the new model of recoverable robust optimization with commitment. More formally, given a combinatorial optimization problem and uncertainty about elements that may fail, we seek a robust solution that, after the failing elements are revealed, can be augmented in a limited way. Importantly, we commit to preserve the remaining elements of the initial solution. We consider different underlying problems and settle the computational complexity of their robust counterparts with commitment. For instance, we show that the reoptimization for the bookings of a vehicle fleet can be solved efficiently.