Online Walking Pattern and Balance Motion for a Biped Humanoid Robot Having a Trunk

Online Walking Pattern and Balance Motion for a Biped Humanoid Robot Having a Trunk

  This paper describes an online locomotion pattern generation for a biped humanoid robot having a three degrees of freedom trunk. This pattern generator uses visual and auditory sensors to detect a walking length and direction from human robot interaction. The motion pattern of the lower-limbs is generated in real time on the basis of walking commands. Depending on the contact information of the landing foot, ZMP is set arbitrarily in a support polygon. To maintain the balance of the whole robot body during walking, the motions of the trunk and the waist are generated by a compensatory motion control method based on the ZMP and the motion of the lower-limbs. To confirm the online locomotion, we have developed a human-like biped robot, WABIAN-RIV that consists of forty-three mechanical degrees of freedom. Versatile biped walking experiments have been conducted on a horizontal plane using visual and auditory information, and the validity of the online pattern generation and the compensatory motion control has been assured.