As an important branch of intelligent robots in the field of industrial manufacturing,spraying robot is widely used in automobile,ship and aerospace and other fields.It plays a very important role in improving the spraying environment and improving the quality of spraying.The traditional spray robot path planning mainly adopts the manual teaching method,which not only poses a certain threat to the health of the teaching staff,but also makes it difficult to guarantee the quality of spraying.When the type of product changes,it has to re-teach.Therefore,it is very necessary to study the automatic path planning method of spraying robot,especially for the flexible production line with mixed production of many kinds of products.Based on the 3D point cloud of different products,this paper introduces an automatic path planning method for the spraying robot.Firstly,the overall scheme of the whole system is designed,including the data acquisition method,pretreatment process and method of obtaining spraying path,etc.We discuss the laser sensor ranging principle and get 3D point cloud by the sensor,perform operations on the point cloud to provide high quality data for path planning algorithm.The pre-processing mainly includes: statistical filtering of point cloud,compression of point cloud data based on edge preservation,calculation and refinement of point cloud normal vector,and the study of improved bilateral filtering algorithm to smooth the small noise of the point cloud.To solve the problems of single sensor acquisition,an acquisition method based on multiple sensors has been designed and the data registration algorithm between multiple sensors has also been studied.Secondly,the model of robot spraying process is established,and various influencing factors in the spraying process are analyzed.Some necessary assumptions are put forward for modeling the spraying process.The law of paint deposition thickness of air spray gun was analyzed,and the linear spray path on the plane and the paint thickness distribution of adjacent paths were calculated and verified by simulation.Based on this,with the uniformity of coating thickness distribution as the objective function,an optimization model of multi-constraint single target is established,and the optimization model is solved and verified by simulation.The optimized spraying parameters will be used in the subsequent spraying path planning algorithm.Then,according to the spraying process,the surface and edge of the workpiece are sprayed separately.An adaptive point cloud slicing algorithm was proposed to select slicing direction according to the characteristics of the workpiece,and the data points obtained by slicing algorithm are sorted according to the requirements of spray path.In order to know the direction of the spray gun at each path point,this paper improves the PCA algorithm and obtains a widely applicable method for direction acquisition of spray gun.The traditional edge detection algorithm is improved to meet the requirements of spraying path planning at the edge of workpiece.Finally,the data fitting method of B-spline curve is studied,and the experiment platform of robot spraying path planning is built.On the platform,we finish real-time acquisition of point cloud,point cloud preprocessing,and verify path planning algorithms based on 3D point cloud.The surface spraying path planning algorithm is validated by the flat workpiece and the free-form surface workpiece.The edge path planning algorithm is verified by the model of a chair and the above two types of workpieces.All the experiments prove the accuracy and rapidity of the surface path planning algorithm and the edge path planning algorithm proposed in this paper. |