Adaptive parameter optimization approach for robotic grinding of weld seam based on laser vision sensor

Automatic robot grinding technology has been widely applied in the modern manufacturing industry. A flexible abrasive belt wheel used to grind the weld can avoid burns on the base material and improve the processing efficiency. However, when the robot grinds a weld seam, the material removal depth does not coincide with the feed depth because of the soft contact and uneven weld height, affecting the weld seam surface uniformity. Given these problems, an adaptive parameter optimization approach for the robotic grinding of a weld seam was proposed based on a laser vision sensor and a material removal model. First, the depth of weld seam removal was obtained by a laser vision sensor based on triangulation in real-time. Then, a macroscopic material removal model considering flexible deformation was established to determine the relationship between the weld height and process parameters, and the model coefficient was experimentally fitted to ensure the accuracy and reliability of the model. In addition, the data of real-time interaction structure between the robot controller and grinding system were obtained and used to unsure that the rotational speed of the belt wheel increased in the convex part and decreased in the concave part, in order to obtain a uniform weld seam surface. Comparative experiments were performed to verify the effectiveness and superiority of the method, and experiments on the surface roughness and weld seam surface height difference were conducted to verify the universality of the method. Experimental results show that the residual height of the weld after grinding can be controlled within 0.2mm, and the maximum removal height difference can be controlled within 0.05mm. The surface roughness Ra of the weld after grinding could reach 0.408 µm.