Heuristic-driven strategy for boosting aerial photography with multi-UAV-aided Internet-of-Things platforms

Abstract

Unmanned Aerial Vehicles (UAVs) are gaining much attractiveness due to the emerging 5G, Internet of things applications and the advances in artificial intelligence. Their application fields encompass both civil and military domains. UAVs can ubiquitously supply many Internet-of-things-driven services; as such, they can be configured into airborne networks to provide flexible aerial views, which is essential for photography and videography-based applications like 3D mapping and real-time monitoring. However, many research challenges need to be tackled to facilitate the deployment of such promising applications. This paper addresses a non-convex NP-hard problem of deploying a fleet of UAVs equipped with rotating gimbal-mounted cameras over large-scaled terrains. The problem is heuristically solved in two phases. Firstly, we introduce a fast parallel multi-verse swarm optimization algorithm. A hybrid multi-objective heuristic coalescing two relevant concepts of stochastic optimization: (1) Improved Multi-Objective Particle Swarm Optimization; (2) Improved Multi-Objective Multi-verse Optimization. This heuristic holds several algorithmic tweaks, such as Pareto-based population splitting, Taguchi-based parameters tuning, adaptive mutation, and is used to derive the most near-optimal hovering coordinates of UAVs under two customized objective functions. Secondly, we adopt an efficient gimbal-based rotation and synchronization strategy allowing the UAVs to rotate their cameras resourcefully in four cardinal directions to boost the aerial photographing coverage with few maneuvers. The claimed performance is rigorously endorsed with intensive simulations and comparative analyses (e.g., analysis of variances, Wilcoxon test, performance index) against twenty multi-objective bio-inspired heuristics. The results show that our approach has the upper hand in terms of efficiency and accuracy.