Capturing the tacit knowledge of the skilled operator to program tool paths and tool orientations for robot belt grinding

This paper describes a novel methodology for generating grinding tool paths, orientations and grinding parameters based on the captured trajectories of a surface finishing tool operated by the skilled operator. First, a variable frame size moving window principal component analysis is performed on the trajectories. Next, features such as the amplitude, speed of the tool centre point, direction of travel, average position, contact force and orientation of the grinder are calculated. Kernel k-means clustering is applied to the trajectory’s feature vectors for segmentation into the tool path primitives. The robotic tool path primitives would be ordered similar to the manual tool path primitives in the grinding strategy. The robotic tool path primitives are then generated based on the boundaries of the manual tool path primitive in a computer-aided manufacturing software. Anchor points are selected from the corners and at intermediate sections along the robotic tool path to inherit the orientations of the manual tool path. Using the inherited skilled operator tool orientations, spherical linear interpolation and spherical spline quaternion interpolation are used to interpolate the orientations for the robotic tool path points along the cross curves, followed by the flow curves, respectively. A finite element model of the belt grinding process is also introduced to quantify the effect of the contact wheel serrations on the material removal rate. This is to account for the difference in contact wheel design between the handheld and robotic grinding tool. The programmed robotic tool paths generated from the trajectories captured from a skilled operator were proven to be able to grind the workpiece to the desired profile within the desired tolerance in a pre-production environment.