Design of an intelligent robotic precise assembly system for rapid teaching and admittance control

•The major contribution is to formulate robot assembly tasks by the proposed machine-learning pipeline of teaching, learning, and control procedures.•We proposed a novel methodology based on GMR and force field to obtain the optimal reference trajectories and expected force and preserve the required curve features for robotic assembly motions by human demonstrations.•We proposed a control algorithm based on admittance to complete robotic assembly tasks when the assembly system had single-axis vibration. In this study, an intelligent robotic precise assembly system for rapid teaching and admittance control was developed. The system comprised three parts, namely the teaching, trajectory learning, and controlling systems. The teaching system controlled the robotic arm through a haptic device and generated tactile feedback for the operator. The trajectory learning system optimized the reference trajectory by mixture Gaussian regression and verified the safety of the reference trajectory through a force field simulation. The controlling system employed admittance control in response to the potential environmental disturbance. A robotic Ethernet-connector assembly system was validated. The results revealed that the arm learned the reference trajectory from ten sets of teaching data. Additionally, the contact force on the aluminium shell in the plugin process was significantly lower than that of the teaching data. Even when the shell was displaced by 1 mm and vibrated at 2 Hz, the robotic arm was able to insert the plastic component in the shell.