关键词: 改进切片法, 深度相机, 点云处理, 轨迹规划, 机器人打磨, 风机叶片

Abstract: With the continuous evolution of wind turbine blade types,the traditional manual teaching and offline programming methods have failed to meet the requirements for obtaining grinding trajectories at low cost and high efficiency. To address the problems such as low operation efficiency and difficulty in matching the model with the actual workpiece in the traditional grinding trajectory planning based on 3D reconstruction, a method for generating grinding trajectory planning for wind turbine blades is proposed by improving the slicing method to process point clouds. An Intel RealSense D455 depth camera is used to collect the three-dimensional point cloud of the blade surface. After preprocessing such as noise removal and improved DBSCAN clustering,the continuous and uniform grinding trajectory is directly generated through the improved slicing method. This method combines Principal Component Analysis (PCA) to achieve coordinate system alignment of the point cloud,and uses bidirectional matching slicing and Savitzky-Golay filtering interpolation to improve the accuracy and uniformity of the trajectory. Simulation results show that the average spacing error of the planned trajectory is less than 8 %,the surface coverage rate is 98.3 %,the smoothness is 0.014,and it compares several mainstream trajectory planning methods for grinding complex curved surfaces,verifying the superiority of this method in terms of time consumption and running efficiency.

Key words: improve slicing method, depth camera, point cloud processing, trajectory planning, robot grinding, wind turbine blade