Research On Tracking Registration And Interaction Technology For AR Flight Training

Flight training simulation technology based on Augmented Reality overlays virtual training information,operation specifications and special tasks on the UAV teleoperation scene in the real world.According to the processing flow of special situations and the state feedback of the UAV,the virtual information can be shown to the pilot accurately in the form of text,model and special effect,as well as voice,to guide them to complete the task.3D registration technology is one of the key technologies of Augmented Reality,which adds 3D virtual objects to real environment seamlessly.Firstly,a registration method based on prior features and Kalman filter fusion is proposed.Recognition of preset artificial signs based on prior color and shape features can improve the recognition accuracy and speed in complex background.The solve Pn P method is used to calculate the camera pose change and get the location of the screen area.On the basis of localization,the Kalman filter method is used to merge the prior feature based vision localization system with the SLAM localization system of AR device itself,the tracking robustness is improved when the observation distance is close and there are few three-dimensional features in the space.Secondly,different display styles are related to the cognitive load and operational efficiency of pilots,so this paper studies the way of displaying information in Augmented Reality environment.based on the cognitive characteristics of pilots,a hint scheme is designed,with visual information as the main information and multi-channel compensation with auditory information when necessary.For visual information,a combination of expression is designed,with text,dynamic effect and models.For the text information,we consider location,color,size and other factors.By doing empirical research,an efficient way of display is proposed.The illumination consistency method based on screen luminance estimation is designed to obtain ambient illumination and simulate screen luminance variation in the form of adding additional light source.So better virtual reality fusion effect can be achieved and visual fatigue caused by luminance difference can be reduced.Finally,this paper evaluates the effectiveness of interaction based on multi-source data.By using the dynamic information of the Staring Point and the feedback information of the operation,the random forest method is used to train the samples,and the classifier of interaction efficiency is obtained.It is applied to the prompting system to optimize the information expression mechanism of Augmented Reality.And a perceptual intensity model is created,which integrates text color,size,space,dynamic effect,model and special effect,assisted with sound information.According to the urgency of the information,the intrinsic value of the information,and the real-time calculation of the interaction efficiency,the perceptual intensity of the information is automatically obtained,and then the display style is determined.