Robotic Multiple Peg-in-Hole Assembly with Deviation Estimation and Fast Insertion Control

Due to the large state-action space and statically indeterminate problem, it is difficult to analyze the robotic multiple peg-in-hole (PIH) assembly tasks, which results in a scarcity of effective and general multiple PIH assembly strategies and control methods. This is particularly evident when faced with situations involving a large number of pegs and holes, as well as small fitting clearances. Besides, when the assembly components are visually inaccessible, there is a lack of efficient methods for deviation estimation, compensation, and adjustment. In order to deal with these problems, this study proposes a general three-stage strategy for robotic multiple PIH assembly which can simplify the planning and control methods. In the first stage, the contact state of the peg-hole is passively adjusted, thereby confining deviations within a plane. Building upon this, the second stage introduces a deviation estimation method that relies solely on force/torque information. In addition, a fast and effective control method for the final insertion stage is presented. Experiments are conducted on three sets of components with six or seven peg-hole pairs. The results demonstrate that the proposed strategy offers effective resolution for the complex multiple PIH tasks. Additionally, the deviation estimation method in the second stage has the ability to greatly compensate for the deviation and control the estimation error within an acceptable range. Lastly, the proposed insertion control method can efficiently complete the assembly in the presence of residual deviations caused by the deviation estimation method.