Abstract
Bodies of multicellular organisms are laid out ac-
cording to morphogens: chemical agents which establish a co-
ordinate system in the early embryo and use this to decide
where body parts should grow. This process offers a mechanism
for the automation of the design of engineering constructs via
evolutionary search in a similar manner to the way biological
evolution has driven the diversity of body forms of life on
Earth. There are many ways of encoding such body plans, but
the main existing approaches have problems around managing
the complexity and stability of the evolutionary search process,
particularly when applied to practical engineering design prob-
lems. This contribution takes the notion of morphogen chemical
gradients and uses it to develop a novel grammar for shape
formation. The central idea is to organise and label the spatial
sub-regions of a design before making decisions regarding the
finished arrangement of structure. This makes it much simpler
to explore compositions of the hierarchy of sub-assemblies in a
design, and to represent design of shape in an evolvable manner.
A worked example of such a morphogenic shape grammar is
described, and used in a multi-objective evolutionary search
for optimal bridge truss structures over four fitness objectives
with two design constraints. The resulting Pareto front shows
a wide variety of bridge designs, demonstrating the power of
this approach to generate a diverse set of viable options to meet
engineering design challenges.
cording to morphogens: chemical agents which establish a co-
ordinate system in the early embryo and use this to decide
where body parts should grow. This process offers a mechanism
for the automation of the design of engineering constructs via
evolutionary search in a similar manner to the way biological
evolution has driven the diversity of body forms of life on
Earth. There are many ways of encoding such body plans, but
the main existing approaches have problems around managing
the complexity and stability of the evolutionary search process,
particularly when applied to practical engineering design prob-
lems. This contribution takes the notion of morphogen chemical
gradients and uses it to develop a novel grammar for shape
formation. The central idea is to organise and label the spatial
sub-regions of a design before making decisions regarding the
finished arrangement of structure. This makes it much simpler
to explore compositions of the hierarchy of sub-assemblies in a
design, and to represent design of shape in an evolvable manner.
A worked example of such a morphogenic shape grammar is
described, and used in a multi-objective evolutionary search
for optimal bridge truss structures over four fitness objectives
with two design constraints. The resulting Pareto front shows
a wide variety of bridge designs, demonstrating the power of
this approach to generate a diverse set of viable options to meet
engineering design challenges.
| Original language | English |
|---|---|
| Title of host publication | IEEE Symposium Series on Computational Intelligence |
| Publisher | IEEE |
| Publication status | Published - 18 Mar 2025 |