Research On Automatic Trajectory Planning Of Spraying Robot

"Made in China 2025" and China's "Industry 4.0" have put forward new demands for intelligent equipment and production,which further promoted the development of industrial robots in China.As an important branch,spraying robots are widely used in automobile,aerospace,ship,furniture and other important fields.The off-line programming technology of spraying robot is mainly applied to the spraying of single-variety and large-batch workpieces at present,For the spraying of multi-varity and small-batch workpieces or non-standard pieces,teach programming is still mainly adopted.Especially in China,the research on spraying robot started late,and the technology of automatic spraying trajectory planning is still immature.Therefore,for multi-variety,small-batch workpieces and non-standard pieces,how to obtain the accurate three-dimensional model and complete the automatic spraying trajectory planning,to achieve more automatic,more intelligent,higher quality spraying is a subject of great academic significance and application value.Based on off-line programming technology of spraying robot,this paper carried out research on the establishment of coating accumulation model,three-dimensional reconstruction of workpiece,optimization of spraying trajectory,as follows:(1)The coating accumulation model of air spraying was established: The coupling method of VOF(Volume of Fluid)+DPM(Discrete Phase Model)was used to numerically simulate the two-phase flow field of air spraying,and the influence of different shapping air pressure on the gas flow field and coating film thickness distribution was analyzed.The static coating accumulation model was established by fitting the coating thickness with the elliptical double ? distribution function,and the reliability of the model was verified by experiments.(2)A new double-nozzle structure for air spraying was studied: Based on the single-nozzle of the air spray gun,a three-dimensional physical model of the double-nozzle was established,and the two-phase flow field of the two nozzles was simulated by using the numerical simulation method.The influence of the spacing L between two paint holes and the angle ? between the axes of the two paint holes on the interference gas flow field and the coating thickness distribution was analyzed,and the optimal structure parameters of the double-nozzle were determined.The simulation results show that the effective coating width and uniformity are better than that of single-nozzle.(3)A semi-dense point cloud acquisition algorithm based on speckle structured light was proposed: Firstly,the distance weighted binarization method was used to preprocess the speckle image.The center pixel coordinates of the speckle were calculated by clustering algorithm,and the speckle was simplified to a single pixel.Then the convolution kernel function was used to create speckle binary descriptor,which not only simplified the speckle descriptor complexity,but also improved the matching rate.The method of calculating Hamming distance was used for speckle image matching,the size of the search block was determined according to the measured distance range,and the matching point pairs were screened from two aspects of similarity constraint and significance constraint.According to the calculation formula of laser speckle depth,the 3D point cloud data can be obtained.Finally,the hardware platform of 3D vision sensor of structured light was built,and the experimental analysis was carried out.(4)A 3D reconstruction method combining structured light method and binocular vision technology for spraying workpiece was proposed.In order to solve the coupling problem of distortion parameters and linear parameters in the existing calibration methods,an improved Zhang's calibration method was proposed,which first corrected the distortion and then solved the linear parameters of the camera.In addition,for the binocular vision system of robot,a calibration method based on single target point was proposed to solve the camera parameters and the robot hand-eye relationship simultaneously.A multi-constraint feature matching method based on ORB was proposed to match images from different perspectives and the 3D coordinates of feature points were calculated.Then,according to the acquired feature points,the transformation matrix between the adjacent frame structure light point clouds is calculated to complete the point cloud stitching.Finally,a 3D reconstruction experimental platform was built to reconstruct the spraying workpiece.(5)The optimization method of spraying trajectory on plane,cylinder and plane-cylinder combined surface was studied,and an optimization method of double-layer spraying planning was proposed: Firstly,based on the coating accumulation model on the plane,the dynamic coating accumulation models on the plane and cylinder were established.Taking the coating uniformity as the optimization objective,the spraying trajectory on plane,cylinder and plane-cylinder combined surface was optimized by using genetic algorithm.For the products with thicker coating or more strict uniformity requirements,an optimization method of double-layer spraying trajectory was proposed.Simulation experiments show that the uniformity of coating is better than that of single-layer spraying.Then,the quaternion method was used to represent the end-pose of the robot,and the smooth trajectory planning of the spraying robot was carried out.Finally,the hardware platform of the spraying robot was built,and the spraying experiment was completed.