Evolving Design Modifiers

Abstract

Evolutionary Developmental biology (EvoDevo) is a process of directed growth whose mechanisms could be used in an evolutionary algorithm for engineering applications. Engineering design can be thought of as a search through a high-dimensional design space for a small number of solutions that are optimal by various metrics. Configuring this search within an EvoDevo algorithm may allow developmental processes to provide a facility to give more immediate, localised feedback to the system as it grows into its final optimal configuration (form). This approach would augment current design practices. The main components needed to run EvoDevo for engineering design are set out in this paper, and these are developed into an algorithm for initial investigations, resulting in evolved neural network-based structural design modifying operators that optimise the structure of a planar truss in an iterative, decentralized manner against multiple objectives. Preliminary results are presented which show that the two levels feedback at the Evo and Devo stages drive the system to ultimately produce feasible solutions.

Enough is Enough: Learning to Stop in Generative Systems

“…Artificial growth in generative systems can vary greatly in implementation, but all must come to a point where they stop growing. This is typically achieved with some hardcoded limit, often even built into the mechanism that generates phenotypes, or as a limit set by a user [14,13,11]. However, to take full advantage of the evolutionary emergence of properties, it may prove helpful to give more control to the evolved generative systems.…”

“…GRNs can be emulated within generative systems through computational structures such as feedforward neural networks and have been successfully implemented for multi-objective problems in [11]. That contribution used a fixed number of development steps and did not investigate the controlled stopping of growth.…”

Investigation of starting conditions in generative processes for the design of engineering structures

“…Price et al [11] have demonstrated the use of Devo processes to create a bracket component. Recent work shown in [12] demonstrated the use of EvoDevo to optimize the structure of the Warren truss problem (bridgelike structure). In this approach, an artificial gene regulatory network (GRN) regulates the growth from an initial structure (seed) to the final solution during the Devo process.…”

“…In contrast to traditional evolutionary processes (where random initial starting points are often chosen), in the EvoDevo process a seed is required to start the process. The concept of a seed, introduced in [12], is defined as an initial solution (structure) whose main requirement is to connect the supports and the external load. In previous work, the seed provided to the EvoDevo process is hand-designed and its quality is relatively good [12] where quality in this context refers to the behaviour of a structure of having a subjective low volume and low strain energy.…”

“…The concept of a seed, introduced in [12], is defined as an initial solution (structure) whose main requirement is to connect the supports and the external load. In previous work, the seed provided to the EvoDevo process is hand-designed and its quality is relatively good [12] where quality in this context refers to the behaviour of a structure of having a subjective low volume and low strain energy. This approach of hand-designing the seed works with the assumption that the designer has the knowledge of what a good quality initial design (seed) looks like.…”