Real-time Grinding System Design for Aluminum Engine Cylinder Head Castings Using Rapid 3D Point Cloud Registration Algorithm

Aluminum engine cylinder head castings and other automotive castings produced through casting must undergo satisfactory grinding before being subject to the subsequent machining process. However, the manual grinding techniques currently are inadequate in satisfying the demands of intelligent manufacturing, as they struggle with product consistency, precision, labor dependency, and time-consuming workstations. In the face of the global environment where new energy vehicles are challenging traditional engine vehicles, coupled with declining population growth, the development and application of intelligent casting grinding technology has become an urgent necessity in the realm of automobile manufacturing. In this research, we concentrate on aluminum engine cylinder heads created from sand mold casting and have developed a comprehensive robotic grinding system for engine cylinder heads, consisting of ’ measurement- machining- quality assessment ’ processes. The system is capable of achieving 3D point cloud data acquisition and processing, extraction of areas to be grinded, full-coverage grinding path planning, and quality assessment based on 3D data. In this paper, a new vertical filtering algorithm is first proposed. By employing a point cloud data registration method rooted in the idea of substitution, the swift registration of scanned point clouds and post-discrete CAD point clouds is realized. When compared to corresponding conventional registration methods, there is a 73.29% reduction in registration errors, and registration time is shortened by 58.72%. Secondly, by incorporating a propagation algorithm into the extraction of areas to be grinded based on the distance threshold method, the extraction accuracy of the areas to be grinded is doubled, reaching 97%. Next, a full-coverage grinding path point planning algorithm based directly on point cloud data is employed, bypassing the complex workpiece reconstruction process and directly extracting grinding path points in the areas to be polished through grid division. Lastly, quality evaluation of the workpieces after grinding is carried out using 3D data. Experiments show that, with the integrated grinding system for complex workpieces based on point cloud data, featuring ’measurement- machining- quality assessment’, the error between post-grinding workpieces and standard workpieces is controlled within 5mm. Moreover, the error is less than 2mm for over 90% of the points, and the remaining margin in the original area awaiting polishing is insufficient by 0.2mm.