| The rapid adoption of Unmanned Aerial Vehicles in industry,agriculture,and the military is not only due to their unique mobility and safety or other application advantages,but also due to the high demand generated by industry development.While Unmanned Aerial Vehicles are widely used,the limited battery capacity and fixed payload have forced researchers to make more improvements,including optimization of the structure of itself,optimization of the mission execution,and planning of the flight path.Path planning technology is one of the key technologies to break through the bottleneck of Unmanned Aerial Vehicles applications.Planning can not only reduce the flight cost,but also provide a certain degree of guarantee in application stability and safety,so one of the important topics of current research and development is to design and improve the path planning methods for Unmanned Aerial Vehicles in mission execution,including single Unmanned Aerial Vehicle scenarios as well as multiple Unmanned Aerial Vehicles scenarios.The specific research work of this thesis consists of three aspects:(1)The Unmanned Aerial Vehicles path planning problem in the context of industrial park monitoring environment is analyzed,then the planning problem is abstracted into a multi-constrained objective planning problem in a multi-obstacles environment.And this thesis solves the problem using the Grey Wolf Optimization algorithm on this basis.Secondly,the algorithm structure is improved by increasing the convergence coefficients to adjust the defects of slow convergence in the late stage,while combining with the Simulated Annealing algorithm mechanism with probabilistic leapfrog nature to ensure that the Unmanned Aerial Vehicle can find the optimal path to reach the predetermined location.Compared with the Grey Wolf Optimization algorithm,the Improved Grey Wolf Optimization algorithm has the 8.3%,16.7%,28.6% and 39.6% lower total cost of path distance on map models with precision of 15,20,25 and 30 respectively.(2)In order to further improve the Unmanned Aerial Vehicle path planning performance and propose the application method of Unmanned Aerial Vehicle clustering,the cooperative exploration planning method of multiple Unmanned Aerial Vehicles is explored and analyzed in the context of environmental monitoring in industrial parks,a quantitative model of Unmanned Aerial Vehicles and environmental information is established,and the path travel strategy module and information exploration strategy module are divided;secondly,the reinforcement learning models of the two modules are created and trained based on the Deep Deterministic Policy Gradient algorithm to accomplish the multiple Unmanned Aerial Vehicles exploration monitoring task.In order to avoid the problem of slow convergence rate of the model,the Artificial Potential Force traction mechanism of the potential field method is introduced into the reinforcement learning model of the path travel strategy module to provide an auxiliary guide to generate path trajectories during the interaction between Unmanned Aerial Vehicles and environment.On the basis of this,this thesis also improves the variation rules of the potential field force traction mechanism to reduce the local optimum problem in the path trajectory generation process.In maps with 21.5%,25.3% and 29.6% of obstacles,multiUAVs could achieve 84.2%,76.7% and 69.9% of environmental exploration by the designed method.(3)An Air Sim simulation platform based on Unreal Engine framework for Unmanned Aerial Vehicles is built,and the path planning methods in actual scenes are applied on the platform,including the improved Grey Wolf Optimization algorithm in single UAV scene and the path planning method of reinforcement learning in multiple Unmanned Aerial Vehicles scene.Considering the conditions and limitations of a series of actual environmental factors,the UAV application monitoring and adaptive planning adjustment in real environment scenes are realized. |
PDF Full Text Request