Vehicle Navigation In The Collaborative Environment Of Drones And Unmanned Vehicles

It is difficult for a single robot to perform tasks in a complex environment.On the contrary,with the strong complementarities between Unmanned Air/Ground Vehicle(UAV/UGV)cooperative systems in many aspects,such as moving speed,load,communication ability and detection precision.Besides,UAV/UGV cooperation systems has applied in many fields,such as military.So UAV/UGV cooperative system has been widely concerned by the researcher.Perception,decision-making and execution are the main scientific issues in UAV/UGV cooperative system.In this paper,the typical search and rescue tasks in UAV/UGV cooperative system are taken,the environmental modeling and vehicle path planning are studied.The main works as follows:Firstly,a hierarchical distributed cooperative control system for UAV and UGV is proposed.Task assignment and information exchange between UAV and UGV through modularization,a communication system between UAV and UGV based on 2.4G Wi-Fi communication network is built by ROS operating system.Secondly,the advantage of UAV's aerial field of view is used.UAV is regarded as "Flying eye" to obtain the information of target and obstacles.At first,the image noise is removed by image preprocessing,image morphology processing.Then,the environment is modeled by SURF algorithm and OSTU algorithm.The results are simulated in a simple environment and a complex environment.Finally,through the change of Euclidean coordinates,the image information acquired by UAV is transformed into the position information that can be recognized by UGV.Thirdly,according to the environmental information acquired by UAV,the advantages and disadvantages of traditional Bidirectional Rapidly-exploring Random Trees algorithm(BRRT),Probabilistic Road Map Method(PRM),Artificial Potential Field method(APF)and A* algorithm are compared in simple and complex environments by using MATLAB.Aiming at the disadvantage of non-smooth A* path planning algorithm,an optimized A* path planning algorithm is proposed and simulated in simple and complex scenarios.The predicted trajectory of UGV reference is given to complete vehicle navigation.Finally,the Donkeycar and Parrot2.0 is selected to build the of UAV/UGV cooperation systems.Besides,the UAV/UGV cooperative communication system is also built.The established experimental platform is used to verify the correctness of the methods involved in this paper and simulations are performed in a typical rescue scenario.Experiment shows that: SURF algorithm and OSTU algorithm can achieve the accuracy,real-time and robustness of target recognition and target edge detection.The optimized A* algorithm can achieve the feasibility and real-time of global path planning.