Vision-guided robotic assembly using uncalibrated vision

Abstract

Many consumer and industrial electronics assemblies are predominantly repetitive manual tasks due to low startup costs and high flexibility. This allows the producer to easily scale back, increase or modify production to respond to the rapidly changing marketplace. However with the rising wages in low-cost countries, and ergonomics and other health concerns, the electronics industries look the automation solution again. The challenge for electronics product assembly automation is how to precisely aligning assembly features on small component for small part tight tolerance assemblies in quickly reconfigurable production. Traditional automation or teach/repeat robotic automation is often prohibitive because of costly fixtures and too time consuming setup and ramp-up. This paper introduces a vision-guided robotic assembly method using uncalibrated vision. The conventional methods of vision-guided robotics in assembly automation solutions (e.g. traditional calibration and visual servoing) cannot perform the assemblies with the grasp error and visually inaccessible assembly features. The proposed vision-guided assembly system based on camera space manipulation (CSM) method, on the other hand, relies on a local calibration method to achieve the high accuracy alignment for the final assembly. Experimental verification is performed to test the accuracy of the alignment, the robustness to grasp error, and easy reconfiguration. Under one camera configuration, the 2D vision-guided assembly robot system successfully assembly the ideal test parts within sub-millimeter tolerance with no manual calibration.