In nature, the morphological changes that occur as cognitive development takes place in human beings and animals have been shown to facilitate learning. Taking inspiration from nature, morphological development aimed at improving learning has been applied in the literature to different robotics configurations and tasks with mixed results. In this paper, we consider a growth-based morphological development approach applied to a quadruped walking task, and, in addition to determining that it improves the learning characteristics, we show that morphological development is something other than just introducing noise as a fitness landscape smoothing technique. Finally, by means of a study of the evolution of the different fitness landscapes, we argue and discuss that learning using growth is equivalent to an incremental learning or fitness landscape shaping approach.
|Title of host publication||Proceedings of the Genetic and Evolutionary Computation Conference Companion|
|Publisher||Association for Computing Machinery|
|Publication status||Published - 2022|
|Event||GECCO '22: Genetic and Evolutionary Computation Conference - Boston, United States|
Duration: 9 Jul 2022 → 13 Jul 2022
|Conference||GECCO '22: Genetic and Evolutionary Computation Conference|
|Period||09/07/2022 → 13/07/2022|