Research On UAV Lane Tracking Technology Based On Vision

With the continuous improvement of the intelligence level of unmanned aerial vehicle(UAV),there are more and more application scenes.Especially in the field of intelligent transportation,UAV autonomous lane tracking can carry out real-time monitoring and early warning of road traffic conditions.In addition,the economic cost of UAV is low,and the application of intelligent technology also makes it easy to operate,which greatly reduces the cost of human resources in traffic monitoring tasks.However,there are still technical problems that have not yet been effectively solved in this field.For example,in the existing UAV lane tracking technology,the algorithm for lane detection has the problem of inaccurate lane detection caused by road environmental factors.Besides,during the flight of the UAV,the failure of the lane tracking caused by influence of external wind field has affected the application of this technology in the field of intelligent transportation to a certain extent.Therefore,in view of the problem that traditional lane detection methods are easily affected by environmental factors,an improved lane detection method based on LaneNet neural network is proposed in this thesis.By improving and optimizing the LaneNet network structure and clustering function,this method enables the network to detect lanes more accurately and reliably.At the same time,in the network training,L1 regularization is used to replace the L2 regularization in the original network,which solves the problem of non-convergence and ensures the stability of lane detection.In addition,in view of the problem of UAV lane tracking that is affected by the wind field,which leads to the failure of lane tracking,a control scheme of UAV lane tracking under wind field interference is proposed.The scheme first established a nominal controller based on PD control on the basis of the UAV dynamic model to analyze and verify the stability of the UAV’s lane tracking control without external interference.Then,considering the influence of external wind field on the UAV lane tracking control,the UAV lane tracking model under wind field interference is established,and the wind field interference observer and the UAV anti-wind interference controller are designed so that the UAV can complete lane tracking flight under the influence of wind field.Finally,a UAV lane tracking prototype system based on vision is designed and implemented from two levels of hardware and software,which is used to implement and verify the method and model proposed in this thesis.