Dinamik hareket primitifleri metodu ile katı olmayan robotik çapak alma makinaları için çevrimiçi yörünge planlama.

Robotic deburring has an advance of precise and repeatable machining process on specific surface profiles. However, different from traditional deburring, deburring robots require generated trajectories on a workpiece with unknown chip thickness, which is a hard task due to the low stiffness of robot arm and abrasive tool used. This study presents a method for planning the online trajectory of a deburring robot by considering tool deflection based on the motion primitives trained from offline trajectories. From offline trajectories, task-related movements of 6-DoF deburring robot and interaction forces between the tool and the known workpiece while performing the deburring process are recorded. Then, by utilizing the laser scanner, the surface of the workpiece is measured after the deburring process in order to determine the difference between the actual depth of cut of the workpiece and given set depth of cut. The remained material is mainly because of the low stiffness of the tool which creates tool deflection. Using force data as perception and remained material as action, Dynamic Movement Primitives method is modified, then trained and used to predict deflection on the tool. The linear deflection and angular deviation compensation are performed in order to make adjustments on the trajectory while performing the deburring operation online. Finally, the form error results of the robotic deburring process with DMP is compared with three other processes; Standard robotic deburring process, robotic deburring with PID control and robotic deburring with Active Disturbance Rejection control.M.S. - Master of Scienc