University of the Aegean
Design, modelling, and experimental evaluation of human order-picking and replenishment systems with AMR-based bin transportation
Abstract
dc:descriptionThe continuous expansion of e-commerce, the increasing complexity of supply chain networks, the growing demand for faster and more accurate order fulfilment, and the need to maintain high levels of customer service have fundamentally transformed the role of modern warehouses. Contemporary warehouse facilities are no longer viewed as passive storage locations but as highly dynamic order fulfilment centers, where speed, accuracy, flexibility, and operational efficiency constitute key sources of competitive advantage. Within this environment, order-picking and replenishment operations represent two of the most critical and resource-intensive warehouse activities, directly influencing operating costs, labor productivity, order lead times, and, ultimately, overall supply chain performance. Order-picking has long been recognized as the most labor-intensive and costly warehouse operation, frequently accounting for more than half of the total operating cost of a distribution center. At the same time, the rapid growth of e-commerce has led to increasingly fragmented customer orders, a continuously expanding product portfolio, and significantly shorter delivery windows. These developments have intensified the need for warehouse systems capable of simultaneously improving productivity, operational flexibility, and service quality. Parallel advances in robotics, artificial intelligence, sensing technologies, and digital manufacturing have created new opportunities for warehouse automation through the deployment of Autonomous Mobile Robots (AMRs), enabling the development of more adaptive and collaborative intralogistics systems. Among the emerging warehouse automation technologies, AMRs have attracted considerable industrial and academic attention due to their ability to automate internal transportation tasks while significantly reducing non-value-added walking activities performed by warehouse operators. Unlike conventional Automated Guided Vehicles (AGVs), AMRs operate without predefined physical guidance infrastructure, allowing dynamic navigation, real-time path replanning, and autonomous adaptation to changing warehouse conditions. These capabilities make AMRs particularly suitable for modern warehouse environments characterized by high product variety, fluctuating demand, and continuously changing operational requirements. Despite their rapid adoption in industrial practice, however, research gaps remain regarding the integrated analysis of manual order-picking activities and AMR-assisted intralogistics operations within a common warehouse environment. Motivated by these challenges, this dissertation investigates the design, analysis, and optimization of an AMR-assisted picker-to-parts warehouse system integrating manual order-picking, replenishment, and autonomous material handling operations. Instead of treating human activities and robotic transportation as separate processes, this research considers the warehouse as a collaborative human–robot system, where warehouse operators and Autonomous Mobile Robots work together to perform order-picking, replenishment, and internal transportation tasks. The main objective is to develop an integrated framework that improves the operational efficiency, economic performance, and long-term sustainability of next-generation warehouse systems. To achieve this objective, the dissertation is organized into two closely interconnected research phases that address complementary aspects of warehouse operations. Phase A focuses on manual order-picking and replenishment activities performed within warehouse picking zones (aisles). Warehouse operators manually retrieve products from storage locations and place them into customer order bins, while replenishment activities ensure continuous product availability at forward picking locations, thereby supporting uninterrupted order fulfilment. Phase B addresses warehouse intralogistics operations performed by Autonomous Mobile Robots. Within the proposed warehouse system, AMRs are responsible for transporting customer order bins between sequential picking zones, delivering completed orders to packing stations, and transporting replenishment inventory from receiving stations to the appropriate warehouse picking zones. Together, these two research phases provide a comprehensive analysis of both the human and robotic components of hybrid warehouse systems, as well as the interactions between them within a unified operational framework. Phase A investigates the impact of alternative order-picking technologies on the operational performance of manual picker-to-parts warehouse systems. Four widely adopted technologies (Pick-by-Paper, Pick-by-RF Scanner, Pick-by-Light, and Pick-by-Vision) are experimentally evaluated under a broad range of operating conditions. The analysis combines controlled laboratory experiments with field experiments conducted in an industrial warehouse environment and adopts a Design of Experiments (DoE) methodology to systematically investigate the influence of multiple operational parameters on order-picking performance. Laboratory experiments provide a controlled environment for analyzing the effects of order-picking technology, picking strategy, the number of order lines, and the number of products per order line. Subsequently, field experiments introduce additional operational complexity by incorporating parameters such as operator experience, SKU storage level, handling unit type, and customer order profile, thereby enabling the proposed methodology to capture the behavior of warehouse operations under realistic industrial conditions. The primary scientific contribution of Phase A lies in the development of an integrated evaluation framework for manual order-picking technologies that combines controlled laboratory experimentation, industrial field experimentation, statistical analysis through Design of Experiments (DoE), Activity-Based Costing (ABC), and environmental assessment based on ISO 14067 standard. Unlike previous studies, which typically examine operational performance, economic evaluation, or environmental impact in isolation, the proposed framework enables the simultaneous assessment of operational, economic, and environmental performance, providing a comprehensive multi-criteria basis for evaluating and selecting order-picking technologies in manual picker-to-parts warehouse systems. The experimental findings demonstrate that the selected order-picking technology constitutes one of the most influential factors of warehouse performance. Among the four technologies evaluated, Pick-by-Light consistently achieves the shortest order completion times and the highest productivity levels. Its high performance is primarily attributed to the elimination of paper handling, the reduction of product search time through visual guidance, and the simplification of product confirmation procedures. Pick-by-Vision also demonstrates significant operational advantages, particularly through hands-free operation and continuous operator guidance, leading to substantial productivity improvements compared with conventional paper-based and RF-scanner-based picking systems. The experimental analysis further reveals that the performance of order-picking technologies cannot be evaluated independently of the operational environment in which they are implemented. Instead, their effectiveness depends on the interaction between technological and operational factors, including operator experience, customer order characteristics, warehouse storage policies, and workstation ergonomics. Consequently, technology selection should not be viewed as an isolated investment decision but rather as part of a broader warehouse design strategy in which technological capabilities and operational conditions are jointly considered. Beyond operational performance, Phase A extends the analysis by incorporating both economic and environmental dimensions. The economic assessment employs the Activity-Based Costing methodology to quantify the actual cost associated with each order-picking technology, while the environmental analysis estimates the corresponding carbon footprint in accordance with the ISO 14067 standard. The results indicate that although Pick-by-Light requires the highest initial investment and exhibits higher energy consumption, its superior productivity leads to the lowest operational cost per order line. Conversely, the environmental assessment reveals that the same technology produces the highest CO₂ emissions per order line, primarily due to its greater energy requirements. These findings highlight the existence of important trade-offs between operational efficiency, economic performance, and environmental sustainability. Taken together, the findings of Phase A demonstrate that the evaluation and selection of order-picking technologies should be approached as a multi-criteria decision problem rather than as a purely productivity-oriented investment. The proposed methodology provides warehouse designers and practitioners with a structured decision-support framework capable of balancing operational, economic, and environmental objectives according to the strategic priorities of individual warehouse systems. While Phase A focuses primarily on improving manual warehouse operations, the second research phase addresses a different but closely related challenge. As customer order bins move between warehouse zones and replenishment inventory is continuously transported throughout the facility, internal material handling increasingly becomes a major cause of overall warehouse performance. This motivates the development of an integrated optimization framework capable of coordinating autonomous transportation activities performed by Autonomous Mobile Robots (AMR) within sequential zone-picking warehouse systems. Phase B therefore shifts the research focus from the manual order-picking activities within an aisle to the optimization of AMR-assisted warehouse intralogistics (movement of customer bins to multiple aisles / picking zones). In contrast to the existing literature, where order-picking and replenishment transportation are modelled as separate optimization problems, the proposed research develops a unified modelling framework that simultaneously considers the transportation of partially completed customer orders between sequential picking zones, the delivery of completed orders to packing stations, and the transportation of replenishment inventory from receiving stations to forward picking zones within the same warehouse system. Within the proposed warehouse environment, AMRs transport customer order bins through a sequential zone-picking process, where each customer order is progressively completed as it moves from one warehouse zone to the next. Simultaneously, AMRs transport replenishment inventory to picking zones while delivering completed customer orders to packing stations. Coordinating these simultaneous transportation flows introduces a highly complex optimization problem involving multiple pickup-and-delivery requests, vehicle capacity constraints, service-time dependencies, soft time windows, battery consumption, and charging decisions. To address this integrated warehouse routing problem, a Mixed-Integer Linear Programming (MILP) formulation is developed at the dissertation, referred to as the Capacitated AMR-assisted Routing Problem with Multiple Pickups and Deliveries, Soft Time Windows, and Charging Constraints (C-AMRRP-MPD-STW-C). The proposed formulation explicitly captures the operational characteristics of AMR-assisted warehouse systems by jointly modelling routing decisions, pickup-and-delivery operations, vehicle capacity limitations, soft time windows, battery state evolution, and charging decisions within a single optimization framework. The primary scientific contribution of Phase B lies in the development of an integrated mathematical modelling framework that simultaneously captures multiple interdependent warehouse operations that have traditionally been addressed as separate optimization problems. More specifically, the proposed model unifies AMR routing decisions, inter-zone transportation of partially completed customer orders, replenishment transportation, vehicle capacity constraints, soft time-window management, battery state modelling, and charging decisions within a single optimization framework, thereby enabling the coordinated optimization of hybrid warehouse intralogistics operations. Unlike conventional warehouse routing formulations, the proposed framework reflects the operational characteristics of sequential zone-picking systems, where customer orders progressively evolve as they pass through multiple picking zones before reaching the packing area. Consequently, routing decisions cannot be considered independently from warehouse operational dynamics, since transportation activities directly influence order completion times, replenishment availability, and overall warehouse performance. The proposed formulation further incorporates conveyor-supported Autonomous Mobile Robots operating in conjunction with stationary conveyor interfaces located at warehouse transfer stations. This system configuration enables fully autonomous loading and unloading operations without requiring direct synchronization between warehouse operators and mobile robots during bin transfer activities. As a result, the proposed warehouse architecture significantly reduces operational dependencies between human and robotic resources while improving transportation continuity throughout the warehouse. An additional methodological contribution concerns the representation of repeated visits to warehouse transfer stations and charging facilities. This is achieved through the introduction of virtual (dummy) nodes, which allow multiple visits to pick up stations, delivery stations, and charging locations without altering the underlying network structure of the routing problem. This modelling approach preserves mathematical tractability while accurately representing the repetitive operational behavior observed in practical warehouse environments. Furthermore, the formulation explicitly models the battery state of each Autonomous Mobile Robot throughout the entire routing process. Rather than treating battery charging as an external scheduling problem, battery consumption and charging decisions are fully integrated into the routing model, allowing energy availability to directly influence route construction and operational feasibility. This explicit representation of battery dynamics provides a realistic description of AMR behavior and enables a detailed investigation of the interaction between routing decisions, charging policies, and operational continuity.To solve the proposed optimization problem, the dissertation investigates three complementary solution approaches. The first employs exact optimization through the developed MILP formulation. The second combines a constructive heuristic with exact optimization, generating high-quality initial solutions that accelerate the subsequent optimization process. The third introduces a hybrid solution framework integrating a Greedy construction procedure, a Large Neighborhood Search (LNS) improvement phase, and exact optimization. The purpose of evaluating multiple solution methodologies is not only to improve computational efficiency but also to investigate the scalability of alternative optimization approaches across different problem sizes and levels of operational complexity. Computational experiments demonstrate that the exact optimization approach can obtain optimal solutions for small-scale problem instances within reasonable computational times. However, as the number of transportation requests and routing decisions increases, the computational burden grows rapidly, limiting the practical applicability of exact optimization for larger warehouse systems. In contrast, the proposed hybrid Greedy - Large Neighborhood Search - Exact framework substantially improves computational efficiency while maintaining high solution quality for medium-scale and large-scale instances. These findings highlight the importance of combining heuristic search strategies with exact optimization techniques when addressing realistic warehouse routing problems characterized by high combinatorial complexity. To further investigate the operational behavior of the proposed warehouse system, the dissertation employs a factorial experimental design to analyse the influence of several critical system parameters, including AMR fleet size, warehouse layout configuration, vehicle travelling speed, and battery capacity. The results identify fleet size as the dominant parameter of overall warehouse performance, while warehouse layout and battery capacity also exert statistically significant effects on system behavior. More importantly, the analysis reveals the existence of distinct operational regimes in which warehouse performance gradually shifts from being primarily constrained by transportation delays to being dominated by routing efficiency as additional AMRs become available. These findings provide valuable managerial insights regarding fleet sizing decisions, warehouse layout selection, and charging policy design for future AMR-assisted warehouse systems. Taken together, the two research phases establish an integrated research framework for the analysis, evaluation, and optimization of hybrid human–robot warehouse systems. Rather than treating manual order-picking and AMR-assisted intralogistics as independent operational problems, the dissertation demonstrates how these complementary warehouse functions can be investigated within a common methodological framework that combines experimental research, sustainability assessment, mathematical modelling, and route optimization. The findings of Phase A provide a systematic methodology for evaluating alternative order-picking technologies by simultaneously considering operational performance, economic efficiency, and environmental sustainability. The results demonstrate that warehouse technology selection cannot be based exclusively on productivity improvements but should instead account for the complex interactions among operational, economic, and environmental objectives. Consequently, the proposed evaluation framework provides warehouse managers with a structured decision-support methodology for selecting order-picking technologies under different operational requirements and strategic priorities. Phase B extends this perspective by introducing an integrated optimization framework for AMR-assisted warehouse intralogistics capable of coordinating multiple transportation processes within sequential zone-picking warehouse environments. The proposed mathematical formulation and solution methodologies provide both theoretical and practical advances for the optimization of hybrid warehouse systems operating under realistic routing, energy, and charging constraints. Beyond demonstrating the applicability of exact optimization and hybrid solution approaches, the computational analysis also provides valuable managerial insights into fleet sizing, warehouse layout design, charging policies, and the operational parameters that most strongly influence overall system performance. From a scientific perspective, the dissertation contributes to multiple complementary research areas, including warehouse operations, order-picking systems, warehouse intralogistics, autonomous mobile robotics, mathematical optimization, and sustainable warehouse design. Its principal contribution lies in integrating methodologies that are typically investigated independently into a research framework capable of analyzing the interaction between human-centered warehouse activities and autonomous robotic transportation systems. By combining controlled experimentation, industrial case studies, economic and environmental assessment, mixed-integer mathematical programming, heuristic optimization, and factorial experimental analysis, the dissertation provides a holistic perspective on the design of next-generation warehouse systems. The proposed methodologies also have direct practical relevance for warehouse designers, logistics service providers, manufacturing companies, and distribution centers seeking to implement hybrid human–robot warehouse systems. The experimental findings support technology selection decisions for manual order-picking operations, while the optimization framework offers practical guidance for designing AMR-assisted transportation systems capable of improving operational efficiency, resource utilization, and service performance under realistic industrial conditions. Overall, the dissertation demonstrates that the successful development of future warehouse systems requires simultaneous consideration of human factors, robotic technologies, operational processes, economic performance, environmental sustainability, and optimization-based decision-making. By bridging manual warehouse operations and autonomous intralogistics within a unified analytical framework, this research contributes to the advancement of hybrid warehouse systems and provides a solid foundation for future research on integrated warehouse optimization, intelligent intralogistics, and collaborative human–robot operations.
Degree
thesis:*- Grantor dc:publisher
- University of the Aegean
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Gialos, Anastasios
- Γιαλός, Αναστάσιος
Subjects
dc:subject × 27- Συλλογή παραγγελιών
- Αποθήκες
- Τεχνολογίες συλλογής
- Αυτόνομα κινητά ρομπότ
- Μεικτός ακέραιος γραμμικός προγραμματισμός
- Αλγόριθμοι δρομολόγησης
- Βελτιστοποίηση
- Υβριδικά συστήματα συλλογής και ανατροφοδοσίας
- Σχεδιασμός πειραμάτων
- Order Picking
- Warehousing
- Intralogistics
- Order picking technologies
- Autonomous mobile robots
- Mixed integer linear programming
- Optimization
- Exact algorithm
- Heuristic algorithms
- Metaheuristic algorithms
- Hybrid order-picking and replenishment systems
- Design of experiments
- Άλλες Επιστήμες Μηχανικού και Τεχνολογίες
- Επιστήμες Μηχανικού και Τεχνολογία
- Μηχανική και Τεχνολογίες, άλλοι τομείς
- Other Engineering and Technologies
- Engineering and Technology
- Engineering and Technologies, miscellaneous
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/62846