Off-line correction method suitable for a machining robotapplication to composite materials

Robotic machining finds its place in a multitude of applications with increasingly restrictive dimensional tolerances. In the machining of left-handed shapes for the production of large composite supports (4-m diameter), the expected shape accuracy is a few hundredths. The industrial robot is not initially compatible with such performance criteria. The literature possesses several ways to improve the accuracy of industrial robots such as stiffness, or stress modeling with dynamic measurement of forces during machining. These methods are difficult to apply in an industrial context because they are too costly in terms of time and investments related to the identification means. This study proposes a new off-line correction based on the mirror correction applied during machining. This method is quickly applicable and required only a 3D vision system. Moreover, it is adapted to any 6-axis serial robot, unlike exiting methods that requires a robot modeling and characterization, which is adapted to a specific robot only. After measuring the position of the tool during a first machining operation, this measurement is compared with the initial program setpoint for identify the robot deviation. A smart and autonomous process is used to re-edit the toolpath to compensate for the deviation. A new machining operation quantifies the correction by producing a part with improved shape tolerances. This article presents the development method, the implementation, and the results obtained following its industrial context. A gain of more than 80% is identified and an analysis of this result is proposed. Future complementary developments are suggested as perspectives.