nach oben

2011 | Buch

Multiagent Coordination Enabling Autonomous Logistics

verfasst von: Arne Schuldt

Verlag: Springer Berlin Heidelberg

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Einloggen, um Zugang zu erhalten

Über dieses Buch

This book describes the implementation of autonomous control with multiagent technology. Therewith, it tackles the challenges of supply network management caused by the complexity, the dynamics, and the distribution of logistics processes. The paradigm of autonomous logistics reduces the computational complexity and copes with the dynamics locally by delegating process control to the participating objects. As an example, shipping containers may themselves plan and schedule their way through logistics networks in accordance with objectives imposed by their owners. The technologies enabling autonomous logistics are thoroughly described and reviewed. The presented solution has been used in a realistic simulation of real-world container logistics processes. The validation shows that autonomous control is feasible and that it outperforms the previous centralised dispatching approach by significantly increasing the resource utilisation efficiency. Moreover, the multiagent system relieves human dispatchers from dealing with standard cases, giving them more time to solve exceptional cases appropriately.

Inhaltsverzeichnis

Frontmatter

Logistics Requirements

Frontmatter

Chapter 1. Introduction

Abstract

Transport, material flow, and logistics have a long history (Jünemann, 1989, pp. 3–10). Many technical inventions enabling logistics date back up to several thousand years (Gudehus, 2007a, p. 6). Transport and logistics lay the foundation for trade if producers and consumers are not located at the same place. Usually, ships have a greater capacity than land vehicles. Hence, it is an age–long practice to employ ships for transporting goods over long distances (Levinson, 2006, p. 16).

Arne Schuldt

Chapter 2. Supply Network Management

Abstract

The objective of logistics is to provide the right quantity of the right objects in the right place at the right time in the right quality for the right price (Jünemann, 1989, p. 18). Its purpose is to provide manufacturing facilities with raw materials and to supply customers with products. Jünemann explicitly points out that minimising costs cannot be the only goal because the other mentioned goals also play an important role in satisfying elaborate logistics demands.

Arne Schuldt

Chapter 3. Autonomous Control in Logistics

Abstract

Supply network management aims at balancing supplies and demands between suppliers and consumers (Section 2.1). This is a challenging task due to the complexity, the dynamics, and the distribution that are inherent in logistics processes (Section 2.3). The autonomous logistics paradigm addresses these challenges by applying local control rather than centralised decision making. To this end, each of the participating logistics entities is itself responsible for satisfying its predefined logistics objectives. Delegating both the autonomy and the ability to make decisions to the logistics objects coincides with the natural distribution observed in logistics. The advantages over previous methods are as follows. Firstly, it is possible to react locally on exceptions. It is thus not necessary to re-schedule the whole system which might even be impossible due to the complexity and the dynamics. Secondly, it is not necessary to reveal internal information and decision processes to a central entity.

Arne Schuldt

Multiagent-Based Approach

Frontmatter

Chapter 4. Agent Technology

Abstract

The autonomous logistics paradigm aims at decreasing the overall problem complexity of supply network management (Section 3.1). To this end, process control is delegated to the participating logistics entities. The question is how this principle can be represented appropriately in a software implementation. Each local entity must itself be enabled to make its decisions. More precisely, the data processing unit of autonomous logistics entities is responsible for decision-making (Section 3.2). This specification is useful for distinguishing data processing from other parts of the logistics entity. The challenging task of implementing decentralised process control, however, is only shifted from the whole entity to one of its parts. In order to prevent ending up with an infinitely nested partitioning, it is important to specify how this unit can actually be implemented.

Arne Schuldt

Chapter 5. Potential for Cooperation in Autonomous Logistics

Abstract

To transform logistics objects in accordance with customer demands, primary logistics functions must be applied (Section 2.1). The applicability of centralised control of supply networks is limited by the complexity, the dynamics, and the distribution of logistics processes (Section 2.3). This finding can be explained by the high number of logistics objects, their manifold parameters, and the dynamic environment. Conventional approaches take a centralised perspective which also requires that all information is centrally available. The paradigm of autonomous logistics aims at overcoming the limitations of conventional control by shifting the perspective to the logistics entities themselves (Section 3.1). These previously inanimate logistics units are provided with logistics objectives by their owners. The entities are then responsible for satisfying their predefined objectives autonomously by requesting execution of the primary logistics functions. Hence, the perspective shifts from individual logistics functions to coordinating all of them.

Arne Schuldt

Chapter 6. Team Formation in Autonomous Logistics

Abstract

Autonomous control delegates decision-making to autonomous logistics entities. The static specification of autonomous logistics networks distinguishes service con- sumers and providers as participants in autonomous logistics processes (Chap. 5). In particular, atomic units such as sales units and individual storage positions are regarded as service consumers and providers, respectively.

Arne Schuldt

Chapter 7. Team Action in Autonomous Logistics

Abstract

The autonomous logistics paradigm envisions that autonomous logistics entities are themselves responsible for achieving their logistics objectives (Section 3.1). Choosing atomic units (Section 5.1) to be autonomous logistics entities leads to a high interaction effort for process control. Hence, there is a potential for cooperation to reduce the interaction effort (Section 5.2). Respective agent interaction protocols facilitate team formation in order to establish organisational structures on demand (Section 6.2). This chapter focuses on the team action step of the model for cooperation (Section 4.3.2). It examines how teams of autonomous logistics entities can actually coordinate their actions for collaborative process control. The third step of the model for cooperation, namely plan formation, is implicitly also addressed here. Pre- defined action schemes are employed which are instantiated for actual inter- action. More complex planning is out of the scope of the interaction-centred focus of this project.

Arne Schuldt

Application and Evaluation

Frontmatter

Chapter 8. Implementing Autonomous Logistics

Abstract

The complexity, the dynamics, and the distribution of logistics processes are major challenges in supply network management (Chapter 2). The paradigm of autonomous logistics addresses these challenges by delegating process control to local logistics entities (Chapter 3). Distributed Artificial Intelligence and particularly agent technology have been identified as appropriate means for implementing autonomous control in logistics (Chapter 4). Chapters 5 to 7 specify a respective agent-based approach. The chapter at hand describes the actual implementation of this specification. Like for all software systems, it is important to test and evaluate the new approach before practical application. Multiagent-based simulation allows testing the behaviour of multiagent systems. In contrast to other kinds of simulation, it reflects the actual system behaviour by directly transferring agents from operation to simulation and vice versa.

Arne Schuldt

Chapter 9. A Case Study in Container Logistics

Abstract

A concept for autonomous control of complex supply networks has been developed in Chapters 5 to 7. This abstract specification has been implemented in Chapter 8. But apart from theoretical considerations, it is also important to examine the application to real industrial logistics processes. Beforehand, it is necessary to investigate the status quo of supply network management in industry. Therefore, a case study has been conducted to examine the current procurement logistics processes of Tchibo (Schuldt, 2006). The following reasons motivate why the logistics of Tchibo is an adequate subject of this case study. Firstly, a high percentage of the suppliers is located in East Asia. As a consequence, the logistics department of Tchibo has to control a complex international supply network. Secondly, Tchibo supplies a great amount of outlets throughout Europe with a weekly changing range of products. This fact underlines the amount of goods that have to be procured and distributed, but also leads to high dynamics in logistics processes. The case study has been conducted in 2006 during a three-month internship at the forward logistics department of Tchibo. It is based on interviews with employees of this department.

Arne Schuldt

Chapter 10. Transition to Autonomous Logistics

Abstract

In autonomous logistics, the participating logistics entities are themselves responsible for achieving the objectives imposed by their owners. Delegating decision-making to the local entities is a significant difference to conventional approaches with centralised control. An operationalisation for autonomous control of logistics processes has been developed in Chapters 5 to 7. The actual implementation with multiagent systems is described in Chapter 8. As a foundation for a transition from centralised to autonomous control, it is important to evaluate the new method. For some aspects, this evaluation can be conducted analytically. Hence, there is no need for simulation in these cases (Wenzel, Weiβ, Collisi-Böohmer, Pitsch & Rose, 2008, p. 15). For more complex runtime interactions of autonomous logistics entities, however, simulation is an appropriate means of investigation. As discussed in Section 8.2, multiagent-based simulation is particularly suited for examining the actual agent behaviour as it would be in real-world operation.

Arne Schuldt

Chapter 11. Conclusion and Outlook

Abstract

This thesis has approached the field of autonomous control in logistics in an interdisciplinary way. Taking the requirements from the field of logistics as a starting point, multiagent technology derived from Distributed Artificial Intelligence has been identified as an appropriate means to implement autonomous logistics. A concept for team formation and team action in autonomous logistics has been developed and investigated exemplarily in a real-world process. This final chapter is intended to summarise findings of the preceding chapters and to draw overall conclusions.

Arne Schuldt

Backmatter

Titel: Multiagent Coordination Enabling Autonomous Logistics
verfasst von: Arne Schuldt
Verlag: Springer Berlin Heidelberg
Electronic ISBN: 978-3-642-20092-2
Print ISBN: 978-3-642-20091-5
DOI: https://doi.org/10.1007/978-3-642-20092-2