Research On The Automated Wall Surface Processing Method Of The Interior Finishing Robot

Due to the huge population,there is a large demand for house in China and the real estate industry is an vital part of Chinese GDP.The interior decoration is the main work to be done before the owner stays in a new house,as a result,many companies specializing in interior decoration have been set up.One of the most important process of interior decoration is the treatment of the rough surface,which is divided into three steps:applying putty,polishing and painting.This work is labor-intensive with few automation technologies involved and has high construction effect requirements.Moreover,building work usually exposes workers in hazardous environment and the conservative construction method usually suffering from low efficiency and uncertainty in quality control.To avoid these problems and reduce the building work duration,a robot system for applying putty on plastered walls and wall surface polishing was proposed in this article.And experiments were implemented to verify the feasibility of the system design.An multi-degree of freedom mobile decoration robot platform for complex interior decoration environment was realized.According to the process requirements of the wall surface treatment and the particularity of the decoration robot construction,the actuators for applying putty and wall surface polishing were specially designed,respectively.With a fully comprehensive investigation of the wall surface treatment process in decoration work,the actuators mechanical structure was determined to meet the stability and construction effectiveness requirements.To carry out motion planning for construction of the manipulator,its kinematics was analysed.According to the actuators design,they was processed and their construction feasibility was verified by experiment.This paper is mainly about the development of the automated construction system for interior decoration robot.The system was composed of an end-to-end artificial neural network to detect the defects on the wall surface and a robotic controller to plan the arm's work trajectory based on the defects visual information.The design of the network includes the collection of dataset,the selection of the network structure,the optimization method,the determination of parameter initialization strategies,the avoidance of over-fitting during training and the application of other learning tricks.According to the defects category and pixel coordinate in image plane obtained by the detection network,a manipulator motion planning controller for continuous autonomous construction was designed.To verify the construction autonomy and effectiveness of the system,the experimental scheme of the robot construction was established and implemented.The comparison experiment was carried out with high-precision three-dimensional structure light camera for wall surface detection to show the advantage of the detection method proposed in this paper.