Multi-Drone Bridge Inspection Path Planning

With the continuous improvement of modernization,people’s material life is further enriched.With the gradual increase in the coverage of the transportation network,the frequency and workload of bridges have been greatly improved compared with the last century.As a basic large-scale building,it plays an indispensable role in transportation.It is a fast,efficient and cheap bridge inspection method.It is an important basis for maintaining bridge health and traffic safety.In recent years,with the rapid development of rotor unmanned aerial vehicle(RUAV)technology and the high degree of freedom of RUAV in space,the bridge inspection problem of RUAV equipped with sensing equipment has become one of the hot research issues at home and abroad.The problem of RUAV collaborative inspection path planning is an important planning task of the inspection process.A reasonable configuration of the RUAV cluster path operation strategy can significantly reduce the cost of the inspection process and improve the inspection efficiency.Therefore,this paper constructs a bridge inspection method for the multiple rotor unmanned aerial vehicle system(RUAVs),and conducts research on the path planning of the multi-UAV collaborative bridge inspection.The environmental model construction and planning method adaptation in the bridge inspection respectively proposed solutions.The main research work is as follows:1.Aiming at the problem of building a bridge environment model,a large-scale bridge point cloud modeling method covering path arc(CPA)is proposed.This method is based on the idea of point cloud slicing,adding supplementary planes and supplementary point sets,and combining the arc length method.The supplementary point set screening process used in the modeling process solves the problem of incomplete modeling coverage.By supplementing viewpoints to filter arcs,it is proposed to use arcs as the smallest planning unit,which reduces the search space in large-scale building coverage path planning,and defines arc generation,merging,and splitting,providing a high-quality and fast planning environment for large-scale building coverage path planning Model framework.2.Aiming at the single RUAV’s path planning problem,single RUAV’s coverage path planning method based on improved genetic algorithm(GA)is proposed.This method is based on the unit decomposition model and mainly solves the problem of how to minimize the cost of the inspection when RUAV is inspected.First,it is proposed to use unit decomposition to establish a bridge inspection path planning optimization model;then,based on the basic model of the heuristic GA algorithm,it is proposed to use graph theory constraints to improve the GA algorithm.The simulation results show that,compared with the enhanced particle swarm optimization(EPSO)and standard GA algorithm,the improved GA algorithm can obtain a more ideal feasible solution.3.Based on the environment model of the covered path arc bridge,the RUAVs system is adapted to the RUAVs system and proposes a RUAVs distributed task allocation method based on the improved extended consistency-based bundle algorithm(CBBA).This method solves the problem of even distribution of RUAVs.On the basis of the original CBBA algorithm,according to the task size,the task splitting and merging adjacency matrix is proposed;in view of the task RUAVs and the characteristics of the split task,a task merging method is proposed,which solves the problem that independent tasks can only be assigned to a single agent in the CBBA assignment process.The feature of evenly splitting tasks also solves the problem of uneven task distribution caused by excessive task set variance.At the same time,the self-organization mapping(SelfOrganization Mapping Net,SOM)algorithm is improved.Based on the SOM algorithm,the mapping of the winning neuron to the data cluster is added,so that the SOM algorithm can be used to solve the generalized traveling salesman problem.This method solves the coverage After the path arc model is assigned by the CBBA task,each RUAV arc set independently quickly plans the problem.