A novel sensing template using data fusion for large volume assembly

Abstract

The size of large components within manufacturing processes leads to complications with automating the processes required to assemble them into larger structures. In recent years, development of multi-sensor networks and breakthroughs in measuring algorithms have allowed for the creation of novel methods of mating large components. One major challenge with deploying sensor networks into production environments is the ability to attach sensors to large volume components. This can be remedied with the use of a sensing template that acts as a pseudo-virtual jig for the assembly process where sensors are embedded onto the template, thus not interfering with the physical assembly. The key step for this sensing template is creating an algorithmic process for accurate component localisation. This paper will introduce an innovative method of using data fusion attached to a sensing template embedded in an aerospace assembly process. A sensing algorithm utilising a Kalman filter allows for accurate component mating with a low error offset and high repeatability. The results of the sensing template show how it is capable of reducing the error offset and improves the repeatability of measurements.

Eytan Canzini

PhD Candidate in Control, Research Scientist in Robotics

Space Systems, Machine Learning, Control & Game Theory