Dynamic Generation and Switching of Object Handling Behaviors by a Humanoid Robot Using a Recurrent Neural Network Model
Kuniaki Noda, Masato Ito, Yukiko Hoshino and Jun Tani
Simulation of Adaptive Behavior 2006 (SAB 2006)
Rome, Italy, 25-29 September 2006
Summary
This study presents experiments on a ball handling behavior learning realized by a small humanoid robot with a dynamic neural network model, the recurrent neural network with parametric bias (RNNPB). Our experiments showed after the robot learned different types of behaviors through the direct human teaching, the robot became able to switch two types of the behaviors as situated to the ball motion dynamics. We analyzed the parametric bias (PB) space to show that each of the multiple dynamical structures acquired in RNNPB corresponds with taught multiple behavior patterns and the behaviors can be switched by adjusting the PB values.