ETD

• admittance control
• assembly in-motion
• dynamic peg-in-hole
• hybrid control
• PBVS

Applications involving the adoption of collaborative robots in modern manufacturing systems with continuous assembly in-motion operations are constantly increasing. In this context, the peg-in-hole task plays a relevant role, although the majority of the existing work relies on the assumption of a static assembly scenario. In contrast, the required tracking and alignment of the parts in dynamic applications make the insertion stage more challenging, especially when tight clearances and additional constraining components are present. The aim of this thesis is to propose a dynamic insertion control strategy applied to an industrial refrigerant fluid recharging system, in which a specific injector, attached to a velocity-controlled robot, has to be connected with a hydraulic quick coupling, moving on an assembly line. In particular, a visual servoing strategy, consisting of an eye-in-hand depth camera, based on a pre-existing 3D pose reconstruction algorithm providing the target relative pose data, is used, in conjunction with a Kalman filter to compensate for camera latency and generate high-frequency state estimates during the approach phase. Once in contact with the moving object, a hybrid force control method is employed to execute an in-motion search strategy to cope with eventual dynamic alignment errors; the final insertion and extraction are, instead, accomplished with a six-DOF admittance controller. The developed moving target tracking phase, together with the subsequent force control algorithm, was validated in a simulation environment and on the physical robot, demonstrating the capability of the proposed method to successfully complete the task despite the challenges posed by strict timing constraints, complex mating part geometries, and a dynamic scenario.