The Dynamic Drone Scheduling Delivery Problem

Giovanni Campuzano*, Eduardo Lalla-Ruiz, Martijn Mes

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

64 Downloads (Pure)


Logistics plays an important role in today’s last-mile economy. Therefore, companies constantly seek for improving their delivery system towards more efficient and sustainable management of parcel distribution. In this paper, we study the Dynamic Drone Scheduling Delivery Problem. The objective is to minimize the delayed deliveries by a fleet of drones located in a central drone station, taking into account the uncertain arrival of parcels, soft time windows, and energy requirements. We develop a Markov Decision Processes (MDP) formulation and solve it approximately by implementing a value-based Reinforcement Learning (RL) approach. We compare our approach with several heuristic dispatching policies and provide insights into the efficiency of our RL algorithm when facing different delivery scenarios.

Original languageEnglish
Title of host publicationComputational Logistics - 13th International Conference, ICCL 2022, Proceedings
EditorsJesica de Armas, Helena Ramalhinho, Stefan Voß
Place of PublicationCham
Number of pages15
ISBN (Electronic)978-3-031-16579-5
ISBN (Print)978-3-031-16578-8
Publication statusPublished - 14 Sept 2022
Event13th International Conference on Computational Logistics, ICCL 2022 - Barcelona, Spain
Duration: 21 Sept 202223 Sept 2022
Conference number: 13

Publication series

NameLecture Notes in Computer Science
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference13th International Conference on Computational Logistics, ICCL 2022
Abbreviated titleICCL 2022


  • Battery charging
  • Drone scheduling
  • Last mile
  • Reinforcement learning
  • UAV
  • 22/4 OA procedure


Dive into the research topics of 'The Dynamic Drone Scheduling Delivery Problem'. Together they form a unique fingerprint.

Cite this