Drone flight time estimation under epistemic uncertainty

Research output: Contribution to conferencePaperpeer-review

Abstract

Drone Logistic Network (or simply, DLN) is an emerging topic in the sector of transportation networks with applications in goods delivery, postal shipping, healthcare networks etc. It is a rather complex system which have different types of drones and ground facilities and it also requires a robust design of the network to ensure optimal time for delivery, efficiency, resilience, risk and cost efficiency along with different other optimizations of ‘Key Performance Indicators’. Moreover, in sectors like healthcare networks, we need to be extra cautious whilst modeling the network as the consequence of failure is severe. Besides these, we also need to work with real-time telemetry data which can be very noisy at times. To deal with the above mentioned technicalities, we propose a robust surrogate modeling strategy through propagation of interval information from the observed data.We are interested in using this surrogate model to simulate contingency scenarios or simply to construct a Digital Twin (DT). For this particular contribution, we are specifically interested in estimating the drone flight time in uncertain conditions. With our proposed method, we obtain interval estimates for our quantities of interest, which can be interpreted as the set of possible values in between the optimistic and pessimistic bounds
Original languageEnglish
Number of pages9
Publication statusPublished - 3 Sept 2023
Event33rd International European Safety and Reliability Conference, ESREL 2023 - Southampton, United Kingdom
Duration: 3 Sept 20237 Sept 2023

Conference

Conference33rd International European Safety and Reliability Conference, ESREL 2023
Country/TerritoryUnited Kingdom
CitySouthampton
Period3/09/237/09/23

Keywords

  • drone logistic network
  • Gaussian process regression (GPR)
  • epistemic uncertainty
  • interval probability

Fingerprint

Dive into the research topics of 'Drone flight time estimation under epistemic uncertainty'. Together they form a unique fingerprint.

Cite this