Robotic design principles emerging from balance of morphology and intelligence

We investigate robotic design principles using an evolutionary robotic system with reconfigurable morphology and intelligence. In this study, we focus on the relationship between an approaching part and the sensor arrangement. In nature, sensors such as the eyes and the nose are concentrated around the approaching part, i.e., the mouth, since the most important task for living creatures is to bring their mouth close to food. Consequently, for living creatures the arrangement of sensors must be based on the position of the mouth. However, a robot does not have a mouth. In addition, an approaching part of a robot can also be changed by a given task. On the basis of this concept, in this study, we investigate the relationship between an approaching part and the sensor arrangement by performing evolutionary simulations. In the simulations, the morphology of a robot is represented as a tree structure consisting of cylindrical cells, and the intelligence of this robot is described as a decision tree. The task given to the robot is to maintain a distance between an approaching part located at the nth cell and an object. The results clarified basic design principles: sensors are optimally arranged on the approaching part and toward the terminal of the tree structure for this task. Furthermore, by analyzing the evolutionary process, we clarified the evolution of a mechanism that creates an effective distance-maintaining function by using the robot´s body structure as a scale.