Both Evolutionary Algorithms (EAs) and Reinforcement Learning Algorithms (RLAs) have proven successful in policy optimisation tasks, however, there is scarce literature comparing their strengths and weaknesses. This makes it difficult to determine which group of algorithms is best suited for a task. This paper presents a comparison of two EAs and two RLAs in solving EvoMan - a video game playing benchmark. We test the algorithms both with and without noise introduction in the initialisation of multiple video game environments. We demonstrate that EAs reach a similar performance to RLAs in the static environments, but when noise is introduced the performance of EAs drops drastically while the performance of RLAs is much less affected.

Moving around in the environment is a fundamental skill for mobile robots. This makes the evolution of an appropriate gait, a pivotal problem in evolutionary robotics. Whereas the majority of the related studies concern robots with predefined modular or legged morphologies and locomotion speed as the optimization objective, here we investigate robots with evolvable morphologies and behavioral traits included in the fitness function. To analyze the effects we consider morphological as well as behavioral features of the evolved robots. To this end, we introduce novel behavioral measures that describe how the robot locomotes and look into the trade-off between them. Our main goal is to gain insights into differences in possible gaits of modular robots and to provide tools to steer evolution towards objectives beyond ‘simple’ speed.

In this work, we evolve phenotypically plastic robots - robots that adapt their bodies and brains according to environmental conditions - in changing environments. In particular, we investigate how the possibility of death in early environmental conditions impacts evolvability and robot traits. Our results demonstrate that early-death improves the efficiency of the evolutionary process for the earlier environmental conditions. On the other hand, the possibility of early-death in the earlier environmental conditions results in a dramatic loss of performance in the latter environmental conditions.

This paper studies the effects of changing environments on the evolution of bodies and brains of modular robots. Our results indicate that environmental history has a long lasting impact on the evolved robot properties. We show that if the environment gradually changes from type A to type B, then the evolved morphological and behavioral properties are very different from those evolving in a type B environment directly. That is, we observe some sort of “genetic memory”. Furthermore, we show that gradually introducing a difficult environment helps to reach fitness levels that are higher than those obtained under those difficult conditions directly. Finally, we also demonstrate that robots evolved in gradually changing environments are more robust, i.e., exhibit a more stable performance under different conditions.

This paper investigates the evolution of modular robots using different selection preferences (i.e., fitness functions), aiming at novelty, speed of locomotion, number of limbs, and combinations of these. The outcomes are analyzed from different perspectives: sampling of the search space, evolved morphologies, and evolved behaviors. This results in a wealth of findings, including a surprise about the number of sampled regions of the search space and the effect of different fitness functions on the evolved morphologies.