The mothership-drone collaborative system is an operational framework in which a mobile platform (the mothership) supports the launch and retrieval of multiple drones, and the drones visit customers under limited payload capacity and endurance constra...
The mothership-drone collaborative system is an operational framework in which a mobile platform (the mothership) supports the launch and retrieval of multiple drones, and the drones visit customers under limited payload capacity and endurance constraints. For efficient mission execution in such a system, collaborative route optimization that jointly considers the routes of the mothership and drones, as well as the launch-retrieval locations, is essential. Many previous studies assume a fixed mothership speed or allow the mothership to stop, and thus do not explicitly account for the minimum and maximum speed bounds that inherently exist for fixed-wing aircraft platforms. This study addresses the Mothership-Drones Collaborative Path Optimization problem that incorporates the minimum and maximum speed bounds of the mothership.
Because the problem requires simultaneously determining launch-retrieval locations and the routes of the mothership and drones in a continuous space, directly optimizing it is computationally challenging. To tackle this difficulty, we discretize the continuous space by sampling a set of candidate launch-retrieval vertices, construct routes by selecting vertices, and finally refine the launch-retrieval locations in the continuous space via nonlinear optimization. For the discretized problem, we formulate a mixed-integer linear programming (MILP) model, which serves as a benchmark that provides exact solutions to the discretized problem. We propose two frameworks with different solution-generation strategies in the discretized space. The metaheuristic-based ISGN uses a genetic algorithm as the core search engine on the sampled vertex set to generate the mothership and drone routes, and then improves the launch-retrieval vertex locations in the continuous space via nonlinear programming. The scheduling-heuristic-based ISHN derives the mothership route by solving a traveling-salesman-problem variant (M1-PDTSP) with the sampled vertex set as input, and subsequently improves the overall routes through continuous space nonlinear programming and local search.
To validate the performance of the proposed frameworks, Monte-Carlo simulations were conducted. As the number of customers increased, both frameworks exhibited increasing objective values and computation times; however, across all scenarios, ISHN achieved a lower mean objective value and a smaller standard deviation than ISGN, confirming its superiority in terms of solution quality and stability. ISHN showed smaller mean and standard deviation in computation time, demonstrating that it consistently produces high-quality solutions within shorter runtimes.