Optimization of robotic drilling operations by leveraging functional kinematic redundancy to minimize joint reversals

Abstract

Industrial robots used in manufacturing suffer from static friction (stiction) and backlash in their joints during joint reversals. The dynamic controller response to these stick-slip and dead-band phenomena leads to error at the end effector, which is especially pronounced and undesirable in precision applications such as aerospace composite drilling. The tolerance requirements for aerospace components are typically smaller than 0.2mm, which is on the boundary of what typical industrial robots can achieve. During robotic drilling operations, even small errors may result in unacceptable tolerances. For this reason, drilling using robots has not been as widely adopted in this sector. Many methods exist to optimally stiffen a robot’s posture, compensate for the anticipated error, or actuate an independently stabilized drilling tool. But these methods do not address the source of the error and often do not result in satisfactory performance. In this thesis, it is shown that by leveraging the functional kinematic redundancy inherent to drilling, the robot can reduce or even completely eliminate joint reversals while achieving the same plunge and retract motions to drill a hole. The rotation about the tool’s redundant work axis is characterized at the start, target, and end positions. The parameter space is searched using Particle Swarm Optimization to converge on the best combination of input parameters which minimize reversals. The proposed methodology is applied to a KUKA KR 6 R700-2 robot with a sample drilling tool, and the performance is analyzed using internal joint position and torque measurements, as well as tool tip position. A reduction in the envelope of the drilling motion of 40% is observed, and the hysteresis commonly seen in robotic drilling motions is significantly reduced.