| Autor: |
Koike, Ryosuke, Ariizumi, Ryo, Matsuno, Fumitoshi |
| Zdroj: |
Artificial Life & Robotics; Nov2022, Vol. 27 Issue 4, p624-631, 8p |
| Abstrakt: |
This paper proposes a novel method to design a robot by simultaneously improving its morphology and controller. The number of rigid parts of the robot and the layout of joints connecting them are represented by a rooted tree, which is called a discrete parameter. Meanwhile, parameters that can be represented by real values are called continuous parameters; these parameters include properties such as the length and the direction of each rigid part, as well as the weights and biases of the controller composed of a multilayer perceptron. For the discrete parameters, we propose an efficient improvement rule, which was established based on the actual evolution of vertebrates. For the continuous parameters, we apply the REINFORCE algorithm. By combining these two methods, we propose a method to simultaneously improve both the discrete and continuous parameters. The advantages of the proposed method are shown by comparison with other design strategies. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink