In the electronics industry, we have different classes of odd-shape insertion machines which assemble THT (through-hole) electronic components on PCBs. However, there remains a group of electronic components which are still assembled manually by workers, such as some transformers, chokes, relays, resistors, capacitors, fuses, connectors and others.
Automated production of this group of components has not yet been possible so far due to the lack of repeatability of their dimensions, possible deviations in the position of the leads, their sometimes relatively large size (up to several cm) and the method of feeding these components – often on trays.
This causes the position of the components picked by the robot gripper from the tray to be random, as the lying components can be shifted/rotated in 3 axes. To insert such a component with the robot arm it is necessary to determine the detailed geometry of the leads including their position and offset in order to insert them precisely in the holes of the PCB and not to damage the PCB and the metallization of the hole or the component itself.
Existing 3D vision systems either did not provide the required measurement accuracy (minimum 0.03 mm in 3 axes) for this task or, with sufficient accuracy, had too long acquisition and processing times (maximum 300 milliseconds required). The precision of the measurement has a direct impact on the speed of the 6-axis robot. In the case of inaccurate measurement, the robot with the insertion algorithm must accurately locate the hole, which takes longer. This results in a longer work cycle for the robot than for a manual insertion process by a worker.