This paper proposes a new three-layer path planning method,where we fused two existing path planning methods(global path and local path)into a single problem for multi-unmanned aerial vehicles(UAVs)path planning for U...This paper proposes a new three-layer path planning method,where we fused two existing path planning methods(global path and local path)into a single problem for multi-unmanned aerial vehicles(UAVs)path planning for UAV.The global-path network layer contains the latest information and algorithms for global planning according to specific applications.The trajectory planning layer represents the kinematics and different motion characteristics,the planningexecution layer implements the local planning algorithm for obstacle avoidance.In the last layer,we propose a new swarm intelligence algorithm called the refraction principle and opposite-based-learning moth flame optimization(ROBL-MFO).In contrast to the classical MFO,the proposed algorithm addresses the shortcoming of the classical MFO algorithm.First,it adapts the moth position update formula to the notion of historical optimal flame average and improves the convergence speed of the algorithm.Second,it utilizes a random inverse learning strategy to narrow down the search space.Finally,the principle of refraction gives the algorithm the ability to jump out of local optima and helps the algorithm avoid premature convergence.The experimental results show that the performance of the proposed algorithm is versatile,robust,and stable.展开更多
文摘This paper proposes a new three-layer path planning method,where we fused two existing path planning methods(global path and local path)into a single problem for multi-unmanned aerial vehicles(UAVs)path planning for UAV.The global-path network layer contains the latest information and algorithms for global planning according to specific applications.The trajectory planning layer represents the kinematics and different motion characteristics,the planningexecution layer implements the local planning algorithm for obstacle avoidance.In the last layer,we propose a new swarm intelligence algorithm called the refraction principle and opposite-based-learning moth flame optimization(ROBL-MFO).In contrast to the classical MFO,the proposed algorithm addresses the shortcoming of the classical MFO algorithm.First,it adapts the moth position update formula to the notion of historical optimal flame average and improves the convergence speed of the algorithm.Second,it utilizes a random inverse learning strategy to narrow down the search space.Finally,the principle of refraction gives the algorithm the ability to jump out of local optima and helps the algorithm avoid premature convergence.The experimental results show that the performance of the proposed algorithm is versatile,robust,and stable.
