Florian Arnold from the University of Antwerp developed an algorithm that enables efficient parcel delivery with cargo bikes. According to Arnold, this is more environmentally friendly and not more expensive than with vans.
Model and algorithms
In his thesis, Arnold Florian from the University of Antwerp has developed models and algorithms to solve such routing problems. Combinatorial optimization problems like routing are usually difficult to solve, and he found that data mining can be a promising tool to obtain more insights into the respective problem structure. A straight-forward next step would be the application of the same approach to rich vehicle routing problems, such as the vehicle routing problem with time windows (VRPTW). The VRPTW constitutes a highly popular problem due to its additional constraints and complexity, and a more thorough investigation of this problem could help to design efficient and simple solution frameworks.
Arnold Florian used the obtained insights to design a heuristic for the VRP which is driven by local search. Many good heuristics for the VRP involve multiple components and parameters, often without offering sufficient motivation for the respective design choices. This situation has improved drastically in the last decade, with the classification of heuristics into meta-heuristics and the rise of standards for testing and tuning.
Integrating inventory and transport decisions
In his thesis he tried to go a step further, and attempted to show that a good implementation and combination of existing ideas is sufficient to obtain a high-quality routing heuristic that can compete with the state-of-the-art. He attempted to build a novel heuristic, to implement new components and frameworks. Routing problems in practice are mostly governed by additional constraints and components.
In his thesis, he had a particular look on how to integrate routing problems with inventory management. Such integrated problems arise in complex supply chain environments in which goods have to be stored and delivered.
Arnold showed that constrains from the inventory side can significantly impact the routing side. In more detailed experiments he investigated the influence of different parameters and used these insights as the motivation to further study the integration of inventory and routing decisions. With a simulation framework, he showed that a tight collaboration between different depots on the routing and the inventory level can elevate the performance of the overall system.