Multi-material Compositional Pattern-Producing Networks for Form Optimisation
CPPN-NEAT (Compositional Pattern Producing Networks and NeuroEvolution for Augmented Topologies) is a representation and optimisation approach that can generate and optimise complex forms without any pre-defined structure by using indirect, implicit representations. CPPN is based on an analogy to embryonic development; NEAT is based on an analogy to neural evolution. We present new developments that extend the approach to include multi-material objects, where the material distribution must be optimised in parallel with the form.
Results are given for a simple problem concerning PV panels to validate the method. This approach is applicable to a large number of problems concerning the design of complex forms. There are many such problems in the field of energy saving and generation, particularly those areas concerned with solar gain. This work forms a first step in exploring the potential of this approach.
KeywordsCPPN NEAT Form Multi-material
Unable to display preview. Download preview PDF.
- 1.Auerbach, J.E., Bongard, J.C.: Evolving complete robots with CPPN-NEAT: the utility of recurrent connections. In: Proceedings of the 13th Annual Conference on Genetic and Evolutionary Computation, GECCO 2011, pp. 1475–1482. ACM, New York (2011)Google Scholar
- 2.Bentley, P., Kumar, S.: Three ways to grow designs: A comparison of evolved embryogenies for a design problem. In: Genetic and Evolutionary Computation Conference, pp. 35–43. Morgan Kaufmann (1999)Google Scholar