Abstract:Task scheduling of multiple UAVs has become a highly active area of research in recent years. Previous research has generally solved the problem in a whole manner, which makes it hard to efficiently generate high-quality task scheduling schemes due to prohibitive computational complexity. By contrast, the paper constructs a novel divide and conquer framework for multi-UAV task scheduling (DCF), which partitions the original multi-UAV scheduling problem into multiple scheduling sub-problems for all the UAVs. To be specific, DCF includes two phases: one is the task allocation phase which produces multiple scheduling sub-problems and the other is the single UAV scheduling phase which generates the scheduling scheme with sequential tasks for each single UAV considering constraints involving UAV capabilities and task demands. Two phases are iteratively performed until the predefined stopping criteria are met. In the task allocation phase, we propose a tabu-list-based simulated annealing (SATL) algorithm to realize task allocation among multiple UAVs. After obtaining the task allocation scheme, a satisfactory scheduling scheme of each single UAV is generated by variable neighborhood descent (VND) algorithm. Extensive experiments and comparative studies are conducted, demonstrating the efficiency of DCF and the proposed SATL-VND algorithm.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the high computational complexity in multi - Unmanned Aerial Vehicles (UAVs) task scheduling. Specifically, traditional multi - UAV task scheduling methods usually try to solve the entire problem at once, which makes it difficult to generate high - quality task scheduling schemes within a reasonable time when facing large - scale tasks and complex constraints. The paper proposes an iterative two - stage optimization method based on the Divide and Conquer (DCF) framework, aiming to reduce the computational complexity by decomposing the original multi - UAV scheduling problem into multiple sub - problems, thereby improving the efficiency of problem - solving and the quality of the scheduling scheme. ### Main Contributions 1. **Proposed a new multi - UAV task scheduling framework based on the divide - and - conquer principle**: - This framework divides multi - UAV task scheduling into two stages: the task allocation stage and the single - UAV scheduling stage. - In the task allocation stage, tasks are allocated to multiple UAVs; in the single - UAV scheduling stage, a sequential task scheduling scheme for each UAV is generated according to the task allocation scheme. - These two stages are executed iteratively until a predefined stopping condition is met. 2. **Developed a Simulated Annealing Algorithm with Tabu List (SATL)**: - In the task allocation stage, the SATL algorithm is used for task allocation. The tabu list mechanism can avoid short - term cycling and repeated visits, enabling the algorithm to diversify solutions and escape from local optimal solutions. 3. **Designed a Variable Neighborhood Descent (VND) algorithm**: - In the single - UAV scheduling stage, the VND algorithm is used to generate high - quality task scheduling schemes considering various constraints. 4. **Conducted extensive experimental verification**: - The experimental results show that the proposed SATL - VND algorithm can not only obtain solutions close to the optimal solution in a relatively short time when dealing with large - scale multi - UAV task scheduling problems, but also outperforms seven other heuristic and meta - heuristic algorithms in terms of solution quality and time consumption. ### Models and Algorithms - **Task Scheduling Model**: - This model is a Mixed Integer Linear Programming (MILP) model, and the objective function is to maximize the total profit of scheduled tasks. - Constraints include: each UAV must start from and return to the base, each actual task has at most one predecessor task and one successor task, flow balance constraints, time - window constraints, flight - range constraints, energy constraints, and memory - capacity constraints. - **VND Algorithm**: - VND is a meta - heuristic algorithm that optimizes solutions by sequentially using multiple neighborhood structures of the current solution. - The initial - solution generation adopts the Highest - Score - First - Allocation (HSFA) algorithm, which scores tasks according to indicators such as task distance, time - window duration, earliest start time, geographical location, and profit, and allocates tasks in the order of high - to - low scores. - Neighborhood structures include the Insert - operator N1 and the Swap - operator N2, which are respectively used to insert unscheduled tasks into the scheduled - task sequence and to exchange unscheduled tasks with scheduled tasks. - **SATL Algorithm**: - It combines the Simulated Annealing algorithm and the tabu list mechanism and is used in the task allocation stage. - It uses the Fuzzy C - Means Clustering Algorithm (FCME) to generate an initial task allocation scheme and iteratively adjusts the allocation scheme to improve scheduling quality. ### Summary This paper effectively solves the high computational complexity problem in multi - UAV task scheduling by proposing an iterative two - stage optimization method (SATL - VND) based on the divide - and - conquer framework, and improves the quality of the scheduling scheme and the solving efficiency.

An Iterative Two-Phase Optimization Method Based on Divide and Conquer Framework for Integrated Scheduling of Multiple UAVs

A resource-constrained distributed task allocation method based on a two-stage coalition formation methodology for multi-UAVs

Cooperative task allocation for heterogeneous multi-UAV using multi-objective optimization algorithm

Coordinated Scheduling of Air and Space Observation Resources Via Divide and Conquer Framework and Iterative Optimization

Multi-objective Variable Neighborhood Descent for Heterogeneous Multi-UAV Coordinated Scheduling

Joint Interdependent Task Scheduling and Energy Balancing for Multi-UAV Enabled Aerial Edge Computing: A Multi-Objective Optimization Approach

DL-DRL: A double-level deep reinforcement learning approach for large-scale task scheduling of multi-UAV

A Dynamic Task Scheduling Method for Multiple UAVs Based on Contract Net Protocol

Multi-UAVs cooperative task assignment and path planning scheme

Reconnaissance and Confirmation Task Planning of Multiple Fixed-Wing UAVs with Specific Payloads: A Comparison Study

Multi-UAVs task allocation method based on MPSO-SA-DQN

Resource Scheduling for UAVs-aided D2D Networks: A Multi-objective Optimization Approach

Cooperative task allocation with simultaneous arrival and resource constraint for multi-UAV using a genetic algorithm

Distributed matching theory-based task re-allocating for heterogeneous multi-UAV edge computing

A Heuristic Task Allocation Method Based on Overlapping Coalition Formation Game for Heterogeneous UAVs

A distributed framework for multiple UAV cooperative target search under dynamic environment

Hierarchical Task Scheduling for Heterogeneous UAVs Based on Hybrid Genetic Algorithm

An efficient multi-objective ant colony optimization for task allocation of heterogeneous unmanned aerial vehicles

Distributed Cooperative Search Algorithm With Task Assignment and Receding Horizon Predictive Control for Multiple Unmanned Aerial Vehicles

Exact and Deep Q-Network Assisted Swarm Intelligence Methods for Scheduling Multi-Objective Heterogeneous Unmanned Surface Vehicles

Multiple Uavs Mission Assignment Based on Modified Pigeon-Inspired Optimization Algorithm