2022/04/02 by Steven Oud, Oud, Steven, Koen van der Pool +1
Computer Science · Engineering · #Evolutionary Algorithms and Applications #FOS: Computer and information sciences #Modular Robots and Swarm Intelligence #Neural and Evolutionary Computing (cs.NE) #Reinforcement Learning in Robotics #Robotics (cs.RO)
paper · pdf · doi:10.48550/arxiv.2204.00934
openalex publication_date 2022/04/02 · openalex created_date 2022/05/05 · openalex updated_date 2026/07/28
The field of evolutionary robotics uses principles of natural evolution to design robots. In this paper, we study the effect of adding a new module inspired by the skeletal muscle to the existing RoboGen framework: the linear actuator. Additionally, we investigate how robots evolved in a plain environment differ from robots evolved in a rough environment. We consider the task of directed locomotion for comparing evolved robot morphologies. The results show that the addition of the linear actuator does not have a significant impact on the performance and morphologies of robots evolved in a plain environment. However, we find significant differences in the morphologies of robots evolved in a plain environment and robots evolved in a rough environment. We find that more complex behavior and morphologies emerge when we change the terrain of the environment.