Entropic optimal transport with congestion aversion Application to relocation of drones

We present a mathematical framework for tempo-spatial entropic optimal transport, motivated by the problem of efficiently routing drones back to logistics centers. To address collision risk, we incorporate a convex penalty term into the transport model. We propose the Sinkhorn-Frank-Wolfe algorithm, a numerically efficient method with theoretical convergence guarantees, and demonstrate its effectiveness through experiments on synthetic datasets. Our approach provides a foundation for optimizing large-scale autonomous drone logistics while ensuring safe and efficient transportation.