Bipedal Robot Walking and Balancing using a Neuronal Network Model

摘要

This paper presents an alternative approach for controlling the walking and balancing of a bipedal robot. The proposed method uses a neuronal network to learn the sensor events obtained via the force sensors and accelerometer and to control the motor events of Bioloid's Dynamixel motors, to walk and balance the bipedal robot. A neuron layer called the controller network links the sensor neuron events to the motor neurons. The proposed neuronal network model (NNM) has demonstrated its ability to successfully control the walking and balancing of a bipedal robot, in the absence of a dynamic model and theoretical control methods.