The Effects of Learning in Morphologically Evolving Robot Systems

Abstract

Simultaneously evolving morphologies (bodies) and controllers (brains) of robots can cause a mismatch between the inherited body and brain in the offspring. To mitigate this problem, the addition of an infant learning period has been proposed relatively long ago by the so-called Triangle of Life approach. However, an empirical assessment is still lacking to-date. In this paper, we investigate the effects of such a learning mechanism from different perspectives. Using extensive simulations we show that learning can greatly increase task performance and reduce the number of generations required to reach a certain fitness level compared to the purely evolutionary approach. Furthermore, we demonstrate that the evolved morphologies will be also different, even though learning only directly affects the controllers. This provides a quantitative demonstration that changes in the brain can induce changes in the body. Finally, we examine the learning delta defined as the performance difference between the inherited and the learned brain, and find that it is growing throughout the evolutionary process. This shows that evolution produces robots with an increasing plasticity, that is, consecutive generations become better learners and, consequently, they perform better at the given task. Moreover, our results demonstrate that the Triangle of Life is not only a concept of theoretical interest, but a system methodology with practical benefits.

Lamarck's Revenge: Inheritance of Learned Traits Can Make Robot Evolution Better

“…Learning allows to fine-tune the coupling between body and brain and also allows for more degrees of freedom that account for environmental changes. The majority of such research consists of applying learning algorithms to the evolvable brains of robots with fixed bodies [ [16][17][18][19][20][21][22][23][24][25] ], but some have also evolved the morphologies [ [26][27][28][29][30][31][32][33][34][35].…”

“…Previous research on artificial evolution combined with learning is mostly limited to Darwinian systems [ 14,32,35,51 ] and the Baldwin effect [ 28,29 ]. The few existing studies on Lamarckian evolution can be divided into three categories.…”

“…Specifically, we apply a Lamarckian system that acts upon the learning layer -what the parents learn can be inherited by the offspring-while solving the reversible genotype-phenotype mapping problem. We compare the Lamarckian system to a Darwinian system [ 35 ] in which learning occurs, but learned traits are not inherited. Both systems include body evolution, brain evolution, and learning.…”

“…Here again, the majority of related work consists of applying learning algorithms to the evolvable brains of robots with fixed bodies [17][18][19][20][21][22][23][24][25][26] . However, it has been argued that morphological robot evolution must include a learning stage immediately after reproduction 27 , and some recent studies have investigated the combination of body evolution, brain evolution, and learning [28][29][30][31][32][33][34][35][36] .Further to being a bio-inspired technique for developing robots, ER forms an alternative approach to studying issues in evolutionary biology. An evolving robot system, be it real or simulated, can be considered as a model of an evolving system of living organisms and used to test hypotheses experimentally 37 .…”

“…Here again, the majority of related work consists of applying learning algorithms to the evolvable brains of robots with fixed bodies [17][18][19][20][21][22][23][24][25][26] . However, it has been argued that morphological robot evolution must include a learning stage immediately after reproduction 27 , and some recent studies have investigated the combination of body evolution, brain evolution, and learning [28][29][30][31][32][33][34][35][36] .…”