Abstract: The inner cavity margin of aero-engine casing castings is unevenly distributed and has a complex structure. To accurately remove the inner cavity margin of the casing, it is necessary to accurately measure the contour of the inner cavity of the casing. It can give full play to the characteristics of the robot's dexterous movement and realize the measurement of complex cavities by using industrial robots with the displacement sensor. However, most of the paths generated by the robot machining tool path generation method consider interference avoidance optimization in the milling, grinding, and polishing and the main concern is the accessibility of the path. Since the angle optimization in non-contact measurement is not considered, it will lead to the problems of larger fluctuations in joint displacement and larger sensor vibration. An interference-free path planning method for robot measurement pose optimization and motion smoothing is proposed. First, the interference-free area for the path point is established with the method of a distance-based interference detection, and the optimization function is established with the acceleration of the robot joint as the performance index which considering the angle between the laser incident direction and the surface normal, the limit of the robot joint. The question is transformed into solving the optimization function by using a piecewise differential evolution algorithm to solve. Experiments are performed on a given path and compared with the graph-based method. The proposed path planning algorithm can effectively constrain the angle between the measurement direction and the surface normal, and can generate smoother robot joint paths compared with methods based on graph search. The running time is shortened by 50.3%, and the instantaneous acceleration of the terminal in the x, y, and z directions is reduced by 15.83%, 30.45%, and 41.72%.

Key words: robot measurement, interference-free area, pose optimization, motion smoothing, differential evolution algorithm