Integrating force and vision feedback for flexible assembly system

Abstract

The Industrial Assembly Challenge at the World Robot Summit 2020 was a competition showcasing the latest technologies in robot assembly tasks. Teams had two days to set up an autonomous assembly system and had to manipulate surprise parts on the competition day. The manufacturing industry has trended towards high-mix low-volume production, which often requires frequent tool changes and can be difficult to set up with industrial robots. Small- and medium-sized enterprises (SMEs) may struggle to use robots for a variety of products, particularly in assembly tasks requiring high accuracy and specialized tools. To improve assembly performance, we propose an assembly method using multiple types of sensors, including an RGB camera and six-axis force-torque sensor. This system allows for flexible assembly tasks without strict alignment, does not require jigs or dedicated hands for each object, and can be easily adapted to different production scenarios. In this paper, we present the motion strategy of our proposed assembly system that participated in the World Robot Summit 2020 and demonstrate how it was able to robustly accomplish tasks such as screw tightening and peg-in-hole by combining information from the camera with information from force-torque sensors.