Optimizing last-mile logistic operations by means of artificial intelligence techniques

The recently presented paper by the University of Deusto presents a route delivery planning and simulation module that forms a core part of the ICT Platform of SENATOR

Researchers from the Deusto Institute of Technology (DeustoTech) at the University of Deusto achieved a groundbreaking milestone by developing a revolutionary software module that utilizes artificial intelligence (AI) to optimize last-mile logistics for sustainable urban delivery. This innovative module, which forms a core part of the ICT Platform of the SENATOR project, aims to enhance the sustainability of cities by developing a new urban logistic model that matching the demand (all the shipments in a day) with the supply (available resources) in order to identify the best and minimal fleet in terms of emissions, restrictions and capability to fulfil the logistic operators commitments by means of the optimum delivery routes.

This research has been presented in the paper entitled “An Artificial Intelligence-based software module for optimizing collaborative delivery in last-mile logistics“, which provides a detailed description of the AI-based optimization method and the architecture and components of the software module.

Addressing the challenges of last-mile logistics

In today’s urban landscape, the proliferation of e-commerce and online retail has made last-mile logistics a critical aspect of modern life. However, the increased freight traffic resulting from deliveries has led to challenges such as congestion, emissions, and road safety, impeding sustainable urban development. To address these issues, the SENATOR project focuses on establishing a smart network operator supported by an advanced ICT Platform to improve decision-making, integration of logistics operators, and planning of logistics operations.

An AI-powered solution

The newly introduced software module tackles the complexities of last-mile logistics through the use of Artificial Intelligence based optimization algorithms. These algorithms enable efficient matching of supply and demand, identification of the most suitable fleet mix, and estimation of optimal delivery routes based on real-time conditions. The module also facilitates last-mile delivery planning across different transport modes, taking into account inter-modality and driving restrictions while optimizing multiple performance indicators.

The software module developed by the Deusto Institute of Technology represents a significant breakthrough in the field of last-mile logistics. By integrating AI and optimization techniques, it revolutionizes the planning, execution, and optimization of deliveries in urban areas.

Optimizing delivery routes in real time

The module employs a hybrid metaheuristic algorithm known as Large Neighborhood Search, as outlined in the research paper. This algorithm provides effective solutions to the complex and well known but not yet resolved Rich Vehicle Routing Problem by incorporating factors such as dynamism (responding to last-minute changes in planning), multi-modality(considering multiple transport modes, like bikes or zero-emission vehicles), inter-modality, multi-objective optimization, and driving restrictions. The developed solution allows the simultaneous optimization of different performance indicators (e.g. distance, time, emissions, etc.) and enables real-time adjustments in delivery planning to accommodate last-minute orders, vehicle breakdowns, and unexpected events.

Extensive testing of the module’s capabilities are being conducted using both real shipment data from Correos in the living lab implemented by the SENATOR project in Zaragoza and simulated data. The tests examined scenarios based on the one hand on the postal operator’s current operations (scenario 1) and on the other hand simulated operation for the implementation of a Low Emission Zone in Zaragoza’s city center (scenario 2).

The tests used three different fleet compositions in terms of fleet electrification in order to understand what impact it may have in environmental and operational terms. The three fleet compositions considered are: current fleet composition in Zaragoza, fleet electrification of around 50%, and fleet electrification of around 100%.

The main conclusions from the two scenarios analyzed with the presented tool are as follows:

  • In scenario 1, increasing the electrification of the fleet leads to a significant reduction in emissions and social costs, with 100% electrification resulting in a 100% reduction in environmental impact. However, there is a slight increase in fixed and capital costs due to the higher price of electric vehicles.
  • In Scenario 2, the results show that fleet electrification has a similar environmental impact reduction as in Scenario 1, but there is an increase in the number of shipments delivered with higher electrification of the fleet due to improved access to the Low Emission Zone.

Driving sustainable urban delivery systems

The introduction of this AI-based software module marks a significant milestone in the SENATOR project’s pursuit of a new urban logistics model. By fostering collaboration between citizens, operators, carriers, and administrations, the project aims to mitigate the negative impacts of distribution in cities, promote sustainability, and enhance overall quality of life. These objectives will be met through a better use and management of logistics operations: sharing resources, promoting the use of non-polluting vehicles in last-mile delivery and improving the use of infrastructure by optimizing delivery routes.

As cities worldwide grapple with the challenges posed by urbanization and evolving consumer behavior, innovative solutions like the SENATOR project’s software module offer hope for the future of last-mile logistics. With further advancements in AI and optimization technologies, the vision of sustainable and efficient urban delivery systems may soon become a reality.

IPIC 2023

Jenny Fajardo from the University of Deusto presented this research at the 9th International Physical Internet Conference (IPIC 2023) on June 13, 2023, in Athens, Greece. The paper showcases the work developed within the SENATOR project framework by the University of Deusto’s partners, including Jenny Fajardo, Antonio D. Masegosa, Pili Elejoste, Asier Moreno, Xabier Cantero-Lopez, and Ignacio Angulo Martinez. ALICE, Alliance for Logistics Innovation through Collaboration in Europe (ETP LOGISTICS), Sense Media Group, with the support of the European Commission, Government of the Hellenic Republic, and ΓΓΕΚ | GSRI organized the IPIC 2023.

For a deeper dive into the findings, you can access the paper here: “An Artificial Intelligence-based software module for optimizing collaborative delivery in last-mile logistics.”


Share on facebook
Share on twitter
Share on email
Share on linkedin


This website uses cookies to ensure you get the best experience on our website. Read Cookie Policy