Component Browsing Based On Hierarchical Navigation Tree And Object Detection In Augmented Reality Scenes

Augmented reality(AR)technology can superimpose computer-generated virtual models into the real world,so observers could see the combination of virtual models and real objects.AR brings technological revolution to the fields of manufacturing,education and training.Microsoft HoloLens is currently mainstream wearable AR device.However,the range of perspective it can provide is limited,and it is impossible to see the entire larger objects when user stands close to them,resulting in a lack of global information.Using augmented reality technology can realize the understanding of the environment.Under the condition of known labeled object,the composition of object can be explained by virtual information.In the absence of specific labeled object,the overall environment can also be analyzed.Under the premise that the labeled object is known,the existing solution considers adding virtual labels to each component of the object in order to clarify the names and positions of the parts,but the way that simply adding labels cannot allow users to grasp the overall structure of the target object,especially for large-scale objects,the lack of global browsing is caused due to the limitation of the hardware device.In the case that there is no specific labeled target,objects in the environment can be identified and located on the AR device,thereby assisting the user to understand the entire environmentIn view of the problem of understanding specific object in AR scenes,this paper proposes the concept of hierarchical navigation tree.By presenting the global structure of the object and the navigation from the two-dimensional interface to three-dimensional object,the specific object is explained.It also improves the negative impact of the field of view.In addition,it adopts multi-channel fusion interaction method to ensure the naturalness of human-computer interaction.In this paper,we take the maintenance training as the application scenario,and use the virtual information to provide the component guidance of the target device,thereby it can speed up the maintenance operation and verify the efficiency of the hierarchical navigation tree.In addition,there are situations where users aim to understand the overall environment rather than focusing on specific object,so this paper also explores intelligent understanding of the real environment in AR scenes.Combined with machine learning model Tiny-YOLO,this paper realizes the interpretation of the real world by object detection and positioning.The main contributions of this paper are as follows:1.Based on the fault tree analysis,the concept of hierarchical navigation tree is proposed in order to meet the functional requirements of maintenance training scenarios and improve the lack of global information caused by narrow field of view.In addition,it can provide efficient human-computer interaction.2.Realizing the detection of real objects based on AR wearable devices,which is convenient for users to understand the surrounding environment.