Research On Path Planning Method For UAV-based PV Inspection System

With its excellent performance and low cost,quad-rotor unmanned aerial vehicle(UAV)has been wildly applied to more and more fields,including inspection of large-scale photovoltaic farm.In a typical scheme,the quad-rotor UAV is equipped xwith an on-board processor,a digital light visible single-lens reflex(SLR)camera and a set of corresponding sensors.After accessing all the pre-defined waypoints,the inspection system perform tasks automatically,get the images of PV modules and finally accomplish the goal of locating defective PV modules and identifying the type of failures and defects by processing these images.However,in practical applications,it is found that if the trajectory planning only aims at reaching the target point without taking the kinematic characteristics of quad-rotor UAV into consideration,the acquisition of useful PV module images can be hardly carried out for fault detection and condition monitoring purposes.In addition,due to the battery life limitation of quad-rotor UAV,the efficiency of the task execution will be directly affected by the result of flight path planning.This paper has studied path planning for quad-rotor UAV oriented to PV inspection.First of all,based on the dynamic model of quad-rotor UAV,this article analyses the impact of flight attitude on the final inspection during whole inspection and establish the model of UAV flight path.Using the multi-objective particle swarm optimization(MOPSO),a series of optional Bezier curves can be obtained as the alternative inspection routes for UAV.The MATLAB simulation result shows that the algorithm can get Pareto frontier effectively,in which we can choose a best one.Off-line UAV path planning based on a pre-defined static map will sometimes cause security problems.For example,we may encounter undefined obstacles during flight.For the reliability requirements of industrial system,we need UAV to monitor the surrounding environment in real time,react quickly to obstacles,and get to the target point on the basis of avoiding obstacles.In this paper,we studied a local path planning scheme based on dynamic window method and carried out the MATLAB simulation result to prove the validity of the algorithm.Finally,we introduce the main hardware configuration and overall system architecture of the UAV inspection system based on Matrice 100.The proposed off-line trajectory optimization algorithm is applied to the system.By analyzing the attitude data obtained from the sensor,we compared it to the previous flight scheme and verified the effectiveness of the method.Then we used the simulation platform under ROS to intuitively picture the obstacle avoidance ability for dynamic window method.