Filming Sport Events with Mobile Camera Drones: Mathematical Modeling and Algorithms

摘要

We introduce and formulate the Sport Event Filming (SEF) problem. We are interested in controlling the movement of a set of camera drones that are filming the event by moving over the field where the event takes place. The objective of the problem is to maximize the satisfaction of event viewers while minimizing the distance traveled by the camera-drones. We model the static version of the SEF problem as a Vehicle Routing Problem with Soft Time Window (VRP-STW), where the whole sequence of actions in the event are known a priori. Since this assumption is unrealistic for real sport events, we propose two families of heuristics to solve the dynamic version of the problem, where the camera-drones do not have any knowledge of the input sequence and move in reaction to the movements of the protagonists of the event. The first family (Nearest Neighbor) is based on a technique used in robotic systems, whereas the second family (Ball Movement Interception) is designed based on specific characteristics of the SEF problem. We present extensive simulation results for both families in terms of average viewer satisfaction and traveled distance for the camera-drones, when several parameters vary.