Acquisition of adaptive behavior for virtual modular robot using evolutionary computation

Authors:
Keisuke Yoneda;Ikuo Suzuki;Masahito Yamamoto;Masashi Furukawa
Affiliations:
Hokkaido University, Laboratory of Autonomous System Engineering, Sapporo, Japan;Hokkaido University, Laboratory of Autonomous System Engineering, Sapporo, Japan;Hokkaido University, Laboratory of Autonomous System Engineering, Sapporo, Japan;Hokkaido University, Laboratory of Autonomous System Engineering, Sapporo, Japan
Venue:
ECAL'09 Proceedings of the 10th European conference on Advances in artificial life: Darwin meets von Neumann - Volume Part I
Year:
2009

Citing 3
Cited 0

Evolving virtual creatures

SIGGRAPH '94 Proceedings of the 21st annual conference on Computer graphics and interactive techniques
Crystalline Robots: Self-Reconfiguration with Compressible Unit Modules

Autonomous Robots
Artificial fishes: Autonomous locomotion, perception, behavior, and learning in a simulated physical world

Artificial Life

Quantified Score

Hi-index	0.00

Visualization

Abstract

In areas such as evolutionary robotics and artificial life, simulating artificial robots and organisms are significant challenges to acquire their proper behaviors that achieve given tasks. This study proposes Animated Robot ("Anibot"), which can behave by obeying physical laws in a virtual 3D environment. Especially, we aim to obtain a control system in evolution which makes it possible to behave "Anibot" autonomously. This paper focuses on a virtual modular robot with a flexible structure and simulating it for learning and controlling. The experimental results show that the modular robot can move toward a light source as its goal in different circumstances. In addition, we discuss an adaptive ability in the different circumstances and a motion mechanism of an obtained behavior.