Designing computational intelligence and data-driven cyber-physical approach for robotic mobile fulfillment system

Abstract

With the rapid development and implementation of ICT, academics and industrial practitioners are widely applying robotic solutions to enhance their business processes and operational efficiencies. A robotic mobile fulfillment system is a mobile robot-based system, generally including the autonomous mobile robot, wireless charging station, mobile storage rack, positioning identifier for location identification, putaway and picking station, and wireless communication infrastructure. The difference between a single-deep traditional warehouse and robotic mobile fulfillment systems is the flexibility of adopting the multi-deep layout. Moreover, the storage location assignment problem and the order picking and packing problem are non-identical as opposed to the traditional human-centric warehouse. This thesis contributes to the theoretical and practical implications in the robotic mobile fulfillment system research field, and extends the application into the manufacturing system. In the first part of the thesis, we address the value creation utilizing cloud-based cyber-physical systems in the robotic mobile fulfillment system. By providing an analysis of cloud services and internet-of-things enhancement, theoretical concepts from the works of literature are consolidated to solve the research questions on how a robotic mobile fulfillment system offering better order fulfillment can gain benefits in terms of operational efficiency and system reliability. Dock grid conflict is a new type of conflict appearing in multi-deep robotic mobile fulfillment systems. Under these circumstances, we address the value creation of robotic process automation under the cloud-based cyber-physical systems in robotic mobile fulfillment systems. A modified A-star algorithm is adopted for calculating the total traveling cost of each moveable rack in the case company layout. Nine common clustering algorithms are applied for the RMFS's zone clustering. The results from the zone clustering are considered as nine scenarios for data-driven order classification to solve the storage location assignment problem. Six common classification algorithms are applied for a detailed comparison which has been conducted with thousands of orders. In the second part of the thesis, we intend to address the value creation of utilizing the industrial internet-of-things driven resource synchronization and sharing-based robotic mobile fulfillment system to enhance the overall operational effectiveness and efficiencies during information transfer and synchronization of resources. A graph theory-based heuristic under the multi-deep robotic mobile fulfillment system is used for computing the shortest path. A model is developed with different storage location assignment rules and strategies under the particular parties to minimize the operation costs. The industrial internet-of-things is enabled resource synchronization and information sharing, and the path is generated under different order scenarios with different algorithms. In the third part of the thesis, we extend the application of a robotic mobile fulfillment system into a manufacturing system. Academics and industrial practitioners are widely considered to enhance manufacturing and operational efficiency and effectiveness assisted with robotic solutions. We intend to develop a cyber-physical production system architecture for tools storage assisted with multi-robots in smart manufacturing and robotic mobile fulfillment system. A decentralized multi-robot path planning is assisted with graph neural networks for adoption in a new proposed smart manufacturing and robotic mobile fulfillment system. We further compare multiple classification algorithms for the mobile robots' action prediction, including a spatial-temporal graph convolutional network.