Degeneracy and networked buffering: principles for supporting emergent evolvability in agile manufacturing systems

  • Authors:

  • Regina Frei;James Whitacre

  • Affiliations:

  • South Kensington, Imperial College London, London, UK SW7 2AZ;University of Birmingham, Edgbaston, UK B15 2TT

  • Venue:
  • Natural Computing: an international journal
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

This article introduces new principles for improving upon the design and implementation of agile manufacturing and assembly systems. It focuses particularly on challenges that arise when dealing with novel conditions and the associated requirements of system evolvability, e.g. seamless reconfigurability to cope with changing production orders, robustness to failures and disturbances, and modifiable user-centric interfaces. Because novelty in manufacturing or the marketplace is only predictable to a limited degree, the flexible mechanisms that will permit a system to adequately respond to novelty cannot be entirely pre-specified. As a solution to this challenge, we propose how evolvability can become a pervasive property of the assembly system that, while constrained by the system's historical development and domain-specific requirements, can emerge and re-emerge without foresight or planning. We first describe an important mechanism by which biological systems can cope with uncertainty through properties described as degeneracy and networked buffering. We discuss what degeneracy means, how it supports a system facing unexpected challenges, and we review evidence from simulations using evolutionary algorithms that support some of our conjectures in models with similarities to several assembly system contexts. Finally, we discuss potential design strategies for encouraging emergent changeability in assembly systems. We also discuss practical challenges to the realization of these concepts within a systems engineering context, especially issues related to system transparency, design costs, and efficiency. We discuss how some of these difficulties can be overcome while also elaborating on those factors that are likely to limit the applicability of these principles.