The Research Of UAV Dynamic Path Planning Based On Optical Flow

With the sustainable development in UAV(Unmanned Aerial Vehicle)area,UAV has been generally applied in traffic order,constructional engineer,mineral exploration,agricultural irrigation and powerline inspection industry.Accompany with the deeply application in automation and machine vision,UAV plays an important role in these areas and helps to save plenty human and material resources.In order to make automation navigation come true,the path planning is the key point and foundation.During the powerline inspection mission,the path planning method is to combine the information between GPS and environment to calculate an obstacle-free path among start point and destination,and helps UAV to finish the mission.This paper aims at doing some researches to provide a clear and optimal path when UAV in powerline inspection.The main works of this paper is shown as follows.The static path planning method: Traditional ACO(Ant Colony Optimal)method is an evaluation algorithm depends on heuristic information.Therefore,the speed and precision of the results are related to the guidance of heuristic.As to traditional ACO path chosen based on probability theory,it makes the method have the problems of premature and local optimal.The APF(Artificial Potential Field)is a method of local obstacle avoidance.Through the potential force between the origin and goal,the improved method calculates the resultant moment and guides the UAV to the destination safely.Therefore,this paper combines the advantage of the APF local optimal and global ACO to improve the ability of path planning method and inherit the ability of robust and global searching ability in UAV.The object detection based on improved optical flow method: Because of the difference between the real environment and simulation in automation navigation mission,it's necessary that UAV needs to equip the vision sensor to explore the surround environment and improve the ability of obstacles detection.With the UAV equipped with fore-vision sensor,this article uses optical flow method to detect the static object.As to the moving of environment,the detection method will follow plenty corners,which will affect the precision of object detection.In order to optimal this problem,the improved method uses the frames difference to separate the object from the environment.Hence,this method uses optical flow to track the separated object and promotes the precision of the optical flow method.The obstacle detection and avoidance based on optical flow method: By monocular distance calculation,the method uses the drifting of optical flow to judge whether the object is a potential obstacle or not.If the object is a threat for UAV navigation,the method will use the laws of optical flow balance to execute the obstacle avoidance order.From the results,it shows that the monocular obstacle avoidance has high precision and speed,can be satisfied with the UAV automation navigation.