Collaborative Logistics Planning and Optimization

Logistics plays a big role to ensure delivery to customers promptly and efficiently. Logistics operators in supply chains may have their own inventory / route optimization engines that operate in silos that produce sub-optimal decisions, causing delays, increased cost and reduced service levels that could have been avoided if proper coordination/consolidation had been sought. Effective logistics planning is needed to enable collaboration and coordination among logistics service providers to provide quality fulfilment services while ensuring cost and carbon efficiency and challenging operational constraints in inventory, routing, packing, and loading/unloading.
The underlying logistics planning algorithms in existing solutions always consider the logistics planning as an optimisation problem and describe it in the form of mathematical programming, which is hard to capture the dynamics and collaboration among players. Furthermore, existing logistics planning solutions always ignore some realistic yet crucial factors such as loading sequence and inventory status. They affect not only the collaboration among the players, but also the eventual logistics cost. Due to the computational hardness, a multi-party logistic planning solution taking into account all above factors is desired yet out of reach so far.
In this project, we will explore to tackle the dynamics (e.g. the demands, inventory, and fulfilment status), and the coordination among logistics players and other players, a learning-to-optimise planning algorithm will be developed. To capture the features of different roles (e.g. customers, fulfilment centers or depots, and fleets), specifically designed deep neural networks will be developed to facilitate the decision making.