Assessing the performance of unmanned aerial vehicle for logistics and transportation leveraging the Bayesian network approach

Applications of drone technology are gradually becoming widespread all over the world. Remote medical support, commodities transportation within due time, live-action movies, cinematography, distant communication support, and many other provisions are provided by drones right now. However, the selection of drones for a particular job is quite sensitive and not all drones are feasible for any job. In this case, we need to scientifically assess and confirm drone performance in logistics and transportation. In this study, we present a Bayesian Network (BN) approach to predict the overall performance of drone technology through four prime criteria (factors), namely: physical specification, technical responses, functional performance, and economic cost. To that end, we have developed a Bayesian network approach to portray the causal relationships between the various factors that affect drone selection based on their performance, and subsequently, predicted the posterior probability of drone performance conditioned upon the aforementioned salient criteria. To evaluate the model further, a number of analyses, such as Bayesian inference algorithm (belief propagation) and sensitivity analyses have been carried out. The results show that when the physical specification criteria are met, the economic factor is most imperious to the overall performance of the drone, followed by functional performance and technical responsiveness. This research would invoke richer dialogue for the researchers and practitioners to select and use suitable drones and develop subsequent policies for different transportation scenarios under uncertainty.