Research And Verification Of Autonomous Navigation Technology For Unmanned System

Unmanned system has attracted more and more attention because of its advantages of "zero casualties" and high intelligence,and autonomous navigation,as an important research field of unmanned system,has gradually become a research hotspot.The primary prerequisite for autonomous navigation of unmanned systems is that no one is involved in the loop,so the ability of autonomous decision-making is crucial.Besides,the environment of unmanned system is complex and changeable,and is much more disturbed in wartime,which demands high precision and stability for the positioning and navigation of unmanned system.Therefore,based on the characteristics of unmanned system,this paper designs a complete autonomous navigation strategy based on the "perception-decision-action" process for unmanned system,and provides a theoretical basis for the autonomous navigation of unmanned system.Firstly,in view of the unmanned system and the complex and changeable working environment,an autonomous decision system based on expert system is constructed,which includes the variable autonomy module,the information acquisition module and the expert system module.At the same time,the autonomous evaluation scheme and standard are put forward to provide the top layer for autonomous navigation of unmanned system.Then the positioning information of the unmanned system is studied,including positioning information,speed and location information.In view of the model error and inertial navigation error in the traditional attitude calculation,the Elman neural network is introduced and the attitude calculation algorithm based on the Elman neural network is proposed,which effectively improves the accuracy and adaptability of the attitude calculation.Aiming at the problem that navigation sensors are prone to failure in complex environment resulting the reduction of accuracy of navigation and the problem that single navigation source can't provide high accuracy position and speed information for unmanned system in complex environment,INS/GPS/SAR/TAN multi source faulttolerant integrated navigation system based on one-class SVM fault detection algorithm is proposed,which effectively improves the accuracy and stability of the unmanned system location information.Next,on the basis of location information,the navigation behavior of autonomous navigation is studied,which is divided into two parts: path planning and obstacle avoidance.In view of the strong and complex randomness of the environment,the PRM algorithm with low complexity and strong real-time is selected as the path planning algorithm for autonomous navigation,and the obstacle avoidance is realized by using the Bug2 algorithm.Finally,the autonomous navigation of the whole unmanned system is simulated and verified.Using Unity3 D as a verification tool,the proposed decision logic,location algorithm and obstacle avoidance strategy are integrated into the simulation environment.The 3D environment is built to simulate the real environment,and a variety of scenarios are set according to the circumstances that the unmanned system may encounter in the navigation task.According to the autonomous navigation behavior under the different scenes,the autonomy evaluation criterion is used to evaluate its autonomy.