A bio-inspired approach for self-correcting compliant assembly systems

Statistical process monitoring and control has been popularized in manufacturing as well as various other industries interested in improving product quality and reducing costs. Advances in this field have focused primarily on more efficient ways for diagnosing faults, reducing variation, developing robust design techniques, and increasing sensor capabilities. However, statistical process monitoring cannot address the need for instant variation reduction during assembly operations. This paper presents a unique dimensional error-compensation approach for compliant sheet metal assembly processes. The resulting autonomous self-correction system integrates rapidly advancing data mining methods, physical models, assembly modeling techniques, sensor capabilities, and actuator networks to implement part-by-part dimensional error compensation. Inspired by biological systems, the proposed quality control approach utilizes immunological principles as a means of developing the required mathematical framework behind the self-correcting methodology. The resulting assembly system obtained through this bio-mimicking approach will be used for autonomous monitoring, detection, diagnosis, and control of station and system level faults, contrary to traditional systems that largely rely on final product measurements and expert analysis to eliminate process faults.