Research On Automatic Spraying System For Lamp Classification Based On Machine Vision

In the field of automobile manufacturing,polycarbonate(PC)materials are often used in lamp manufacturing.PC plastics are subject to smashing and yellowing during driving,due to their poor temperature resistance and hardness.As a result,the lighting effect affects the safety of the drivers.Therefore,before the lamp leaves the factory,a layer of hardened paint is applied on the surface to make the illumination of the lamp improved.For the characteristics of small batches and various types of lamp,the traditional spraying method can not meet the producer's pursuit of efficiency.In order to solve this production problem,An automated spraying system of auto lamps based on machine vision is designed with the support of the National Natural Science Foundation of China(61503162,51505193)and the Jiangsu Natural Science Foundation of China(BK20150473)and taking the project named ‘Auto Spraying of Vehicle Lamp' of Jiangsu Times Robot Science and Technology Ltd as the platform.The main work of this paper is as follows:Firstly,according to the characteristics of the lamp workpiece,the overall scheme of automated spraying system based on machine vision for the lamp is designed.The whole system is divided into three parts: conveyor system,vision system and spraying system.The selection of the various modules including the camera,light source,industrial robot and spray gun,as well as the design of the spray process were completed.Secondly,the method of auto lamp recognition based on vision module was studied.Due to the influence of noise in spraying environment,the denoising method of auto lamp image is studied.Aiming at solving the insufficient problem of data set,image enhancement methods are studied.Also the corresponding relationship between image pixel coordinates and geodetic coordinates is determined by camera calibration.The visual recognition method based on Convolutional Neural Network(CNN)is used for classification of auto lamps.An improved Faster R-CNN model for auto lamps recognition is proposed.Four different types of auto lamps are classified and located by using TensorFlow framework.Thirdly,the trajectory optimization method of the lamp spraying is studied.The cumulative rate model of parabolic coating thickness was established,and a slicing algorithm based on car lamp spraying was proposed.The width and speed of the two adjacent spraying paths were determined by taking the deviation of the overlapping area of coating thickness from the ideal thickness as the optimization objective.An improved Particle Swarm Optimization(PSO)algorithm is proposed for the optimal combination of all spraying trajectories.Fourthly,a human-computer interaction interface based on PYQT is designed.The trajectory selection of spraying is determined by identification and location of the auto lamp.The layout of the workstation of the automatic spraying system is completed on Robot Studio platform,and the simulation of spraying trajectory is realized on Robot Studio for one type of car lamp.The feasibility of the whole automatic spraying system is verified by simulation.