Robotic Assembly Strategy With Wrist Force Sense for Narrow Clearance Peg-in-Hole

Abstract

Narrow clearance peg-in-hole assembly is a typical industrial assembly scene. Due to the motion accuracy limitation of industrial robot system and the uncertainty of the assembly circumstance, robotic autonomous assembly of narrow clearance peg in hole assisted by machine vision and force feedback has always been a research focus in this field. However, few studies focused on accomplishing it in a short time while maintaining minimal forces. On the other hand, in some extreme scenarios, such as nuclear, camera lenses may age quickly when exposed to high-dose radiation, requiring a non-visual peg-in-hole method. In this article, aiming at peg-in-hole assembly tasks without vision, and inspired by wrist force sense of human, a robotic narrow clearance peg-in-hole location and insertion methods with wrist force sensation are proposed. The collaborative robot obtains the accurate location of the hole by touching the hole edge several times, named “force-sense-based location”. And then, the peg is automatically inserted into the hole by the collaborative robot with four basic motions named “approach”, “slippage”, “rolling” and “drilling”. A geometric theoretical model was deduced to guarantee the feasibility of the developed method. To avoid excessive contact force, all four motions were designed with admittance hybrid/position control in different directions. The experimental results demonstrate the effectiveness of the proposed method, while the key factors affecting their performance were analyzed. Finally, we compared the methods with three peg-in-hole methods and comprehensively analyzed their performances in terms of assembly time, assembly force, and sensitivity to initial bias. This article provides a new effective method for automatic assembly of narrow clearance peg-in-hole tasks.Note to Practitioners—The efficiency of product assembly has significantly increased thanks to the emergence of robotics. Robots are progressively taking the role of humans in many manual assembly procedures. Peg-in-hole assembly is a common assembly activity accounting for about 40% of the total assembly work. The majority of peg-in-hole assembly techniques start by locating the holes visually, then make modifications based on high-frequency force input. However, vision is frequently prohibited in many real-world assembly scenarios, such as those that actually happen in the dark, low light, or radiation. This paper proposed the use of a UR5e collaborative robot mounted with a six-dimensional force/torque (FT) sensor to complete the peg-in-hole assembly task. The collaborative robots are excellent for applications requiring frequent interactions such as peg-in-hole assembly for high security and great price-performance ratio. A robotic peg-in-hole locating algorithm based on touch sense is presented. This method does not require the assistance of visual equipment like cameras or laser-type displacement transducer. Instead, the location and contact force of the component in various contacts are used to estimate the part’s positional information. The presented method can locate different rigid cylindrical parts and performs fairly precise positioning. The algorithm’s tolerance for position error uncertainty allows it to handle chamfered peg-in-hole assemblies with a narrow clearance ( $25\sim 90\mu $ m). The mathematical derivation ensures the reliability and adaptability of the assembly approach. But a limitation of this approach is that it can only presently be used for cylindrical peg-in-hole assembly with flat end faces. In the future, to expand the application of our method, we will carry out theoretical derivation, control strategy design for more parts with typical shapes, as well as the placement and insertion task of irregular parts.