Joint or decoupled optimization: Multi-UAV path planning for search and rescue

Abstract

In this paper, we focus on path planning of drone teams deployed for search and rescue missions. The goal of the mission is to detect a target, inform the rescue personnel at the ground base station (BS), and form a communication relay chain between the target and the BS as fast as possible. Such missions where both detection and connectivity requirements need to be met can be planned by formulating (i) a single objective optimization problem with connectivity constraints; (ii) a multi-objective optimization problem where mission and connectivity needs are jointly optimized or (iii) mission and connectivity tasks are optimized decoupled from each other. Both joint and decoupled approaches have merit in terms of mission times, connectivity, cost and/or implementation. In this paper, we compare selected joint and decoupled multi-drone path planning approaches from mission and connectivity perspectives. We illustrate the trade-off between performance metrics from both viewpoints and show that depending on the available resources (e.g., number of drones) and the search area most suitable planner can change. We then propose a hybrid planner that utilizes joint optimization for the search drones and decoupled optimization for the relay drones. Hence, the proposed scheme has a flexibility due to allowing different search path planners to be used and a connectivity-wise better pre-mission plan. Our analysis shows that the hybrid scheme results in a better connectivity and total mission time if there are enough drones, but for very small number of search drones, hybrid scheme leads to a higher mission time than the joint scheme due to reservation of drones for relaying purposes.