Research And Implementation Of Real-time Detection Control And Visualization In Wall Plastering

As the population growth rate continues to decrease and the labor force decreases,all walks of life seek to improve productivity through the widespread use of robots,especially the construction industry,The construction industry is a very old industry and accounts for a large proportion of the national economy.However,due to the high uncertainty and complex construction environment in the construction industry,the degree of danger is second only to the mining industry.The current construction industry is still dominated by traditional workers’ decoration,which makes it difficult to improve production efficiency.The low degree of automation,low construction efficiency,long cycle and high cost are all key factors restricting the development of the interior decoration industry.In order to solve the above problems,this thesis studies the automatic implementation of wall plastering in interior decoration robot,including the real-time detection and control in the process of wall plastering and the implementation of wall quality detection system after plastering.When the indoor plastering robot is used for plastering,it is very important to determine the plastering method according to the flatness of the wall to be plastered.The concave convex,defects and other factors on the wall may lead to large errors in the flatness of the last plastered wall.Therefore,the robot can not only use a separate method for plastering the wall in the plastering process,but adjust the plastering strategy in real time according to the situation of the wall.Based on the above problems,this thesis mainly makes research in the following aspects:1.The overall wall plastering scheme of the indoor plastering robot is designed.The scheme comprehensively considers various problems in the wall plastering work,and constructs and designs the entire wall plastering platform.Not only the hardware platform including robot mobile platform,mortar supply facility,depth camera and other equipment is designed,but also the overall software implementation plan including wall plastering detection control scheme and wall quality visual scoring model.2.A real-time detection and control system for indoor plastering robot wall plastering work is designed.The system consists of a real-time detection system based on an active vision-based depth camera and a sensor-based mechanical construction controller.The system mainly acquires wall information through the depth camera,and performs segmentation and image processing on the acquired wall information.Distinguish the mortar information in the plastering process and the wall surface information to be plastered,and calculate whether the current amount of mortar is overflowing or lacking,so as to control the change of the mortar supply rate of the plastering machine.3.A visual scoring model of wall quality based on deep learning is proposed.The model classifies the picture information of the plastered wall by deep learning,and divides the wall pictures into flat wall pictures and non-flat wall pictures.After each construction is completed,the overall wall picture is divided and scored,and finally the wall quality score is fed back to the construction personnel,and the wall quality is evaluated through a visual method.