Research On Path Planning And 3D Reconstruction Based On Indoor Scenes Exploration Vehicle

In recent years,with the increase of various types of indoor emergencies and dangers,using intelligence vehicle as an indoor environment exploration vehicle for indoor environment three-dimensional reconstruction and assisting rescue workers to carry out indoor rescue are gradually becoming the focus of recent research.However,the current research on indoor rescue vehicles mainly focuses on the indoor reconstruction algorithm itself,and the research on efficient collection of reconstruction information is limited.The combination of reconstruction information collection method and path planning method can effectively reduce the time of rebuilding information collection and improve the rescue efficiency.The key and difficulty of fast three-dimensional reconstruction of indoor environment by indoor rescue vehicle is information collection viewpoint and viewpoint path planning,which has important practical significance to study the related technologies involved.Firstly,the software and hardware platform and system model of indoor rescue vehicle are designed and built.According to the characteristics of crowded and narrow indoor rescue environment,combined with the structure and operation characteristics of rescue vehicle,and the features of three-dimensional reconstruction information collection,the hardware structure of indoor rescue vehicle that can adapt to narrow and crowded environment and move relatively flexibly in the environment is designed,and its software is designed.The sensor parameters applied in this paper are analyzed briefly,and the mathematical models of indoor rescue vehicle and camera are established.Finally,the representation method of indoor map is analyzed and explained.Secondly,an interior 3D reconstruction view planning method based on point cloud semantics segmentation is presented.A mathematical model for sensor collecting indoor reconstruction information is established,and the scope of sensor collecting information is determined.A viewpoint planning method for indoor rescue environment is proposed based on point cloud semantics segmentation for the presence of important and nonimportant people and things in the room.The semantics segmentation method is used to mark and separate point clouds representing important people and things.The point cloud representing important people and things is marked and separated by semantics segmentation method.In the process of point cloud simplification,OBB surround box method is selected as the standard to measure whether the simplified point cloud can replace the origin cloud.The algorithm is verified by simulation experiments of view planning.Then,the path planning algorithm in the process of indoor perception of rescue vehicle is studied.In order to enable rescue vehicles to traverse each viewpoint,the traveler problem is studied.Based on the principle of ant colony algorithm,a path planning algorithm with the shortest path traversing all viewpoints is implemented.To meet the obstacle avoidance requirements during operation,based on the principle of DWA algorithm,effective obstacle avoidance is realized.The viewpoint traverse algorithm and obstacle avoidance algorithm are verified by the viewpoint traverse simulation experiment and the obstacle avoidance simulation experiment by using customized samples and simulating indoor environment.Finally,the positioning method of rescue vehicle is studied with indoor map.Based on the analysis of indoor environment characteristics and grid map updating methods,an optimization method of grid map is proposed to eliminate the defect of grid map offset and improve the positioning accuracy by combining SLAM method and feature point extraction method.It is validated by simulation experiments which simulate indoor environment.