| Environmental perception systems have long been a research hotspot in the field of intelligent mobile devices and virtual reality(VR)/augmented reality(AR).These include high-precision map reconstruction,robust context awareness,and accurate path planning.With the development of related technologies such as 3D reconstruction and path planning,unmanned vehicles can explore unknown and complex terrain,indoor scenes can rely on the positioning and navigation of intelligent mobile devices,and the research of reconstruction systems and environmental perception systems in VR/AR scenes is becoming more and more important.However,at present,the environmental perception of virtual or real scenes in VR/AR application scenarios faces problems such as high sensor costs and high costs to generate complex or unknown terrain.Based on indoor and outdoor real and virtual scenes,this paper combines the unknown and complex operating environment of unmanned vehicles in VR application scenarios,and the characteristics of intelligent devices in AR scenarios that need to interact with dynamic scenes when facing environments with single characteristics.In view of the problems existing in the existing method,the following work has been carried out:(1)Construction and verification of environment perception system in VR scene:Unmanned vehicle systems often need to explore unknown environments,so it is critical to verify that the system used to explore unknown targets has robust situational awareness in VR scenarios.Aiming at the problems of unknown and complex operating environment of unmanned vehicles in application scenarios and cumbersome verification process,this paper proposes a context awareness verification system in VR scenes,which can verify(even if Simultaneous Localization And Mapping(SLAM)system is aware of the environment in complex scenes.This paper simulates optical cameras and depth cameras commonly used in robotics,intelligent driving and other related fields,and ensures the time synchronization and spatial synchronization of various sensors in VR scenes.In addition,the interaction model between the unmanned vehicle and the scene is also established.Finally,two comparative methods are used to verify the effectiveness of the proposed system.(2)AR environment perception system to build a computer room operation and maintenance platform:In scenarios such as computer room operation and maintenance where smart mobile devices are worn for operation and maintenance,mobile devices often need to face realistic scenarios such as townhouses and large-area walls,which have the characteristics of single features and difficult to extract effective features.It is difficult for mobile devices to carry complex sensors and need to take care of other tasks running in the device,so it is difficult to maintain accurate positioning for a long time;In addition,in order to maintain real-time performance,mobile devices cannot output highprecision maps,and high-precision maps are the basis for accurate navigation and device positioning.In order to solve this problem,this paper designs a lightweight environment perception system that can be installed on smart mobile devices,which applies remote offline mapping and two-dimensional code repositioning and other schemes,which can not only ensure stable positioning and highprecision map reconstruction results in larger scenes,but also ensure the performance of other daily operation and maintenance tasks in the system,improving the user’s augmented reality experience.In addition,the daily operation and maintenance task module of the system also integrates functional modules such as expert remote annotation based on scene positioning and accurate map reconstruction methods,real-time live broadcast of operation and maintenance scenarios,operation and maintenance scenario model management,real-time navigation of operation and maintenance scenarios,and electronic fences.Finally,the evaluation confirms that the environment perception module integrated in this system can meet the real-time interaction needs of frontline personnel in scenarios and remote experts in O&M scenarios.(3)Passability analysis of multi-level integration map of VR/AR scene:Whether in VR or AR applications,digital 3D scenes need to be processed in the system,and digital 3D models are complex and dynamic.In response to this problem,this article divides the surrounding scene into passable terrain,non-passable terrain,and specific targets with detection value.According to the environment map and the movement ability of the moving subject,the interaction model between the moving subject and the environment is established.Finally,tens of thousands of random paths are generated in the simulation environment and their passability is verified,that is,the moving body can complete two-way passage in each random path,then the path seeking is considered to be successful,and the proportion of the number of successes is used as the passability probability after counting all the results.In addition,the comparison verification is completed in the real scene. |
