Optimizing Upper Body Motion of Humanoid Robot

Recently there are various kinds of research on humanoid robot, such as motion control, man-machine interface, visual or haptic feedback, artificial intelligent, and so on. For motion control, the present works still have many problems since the number of degrees of freedom (DOFs), range of joint motion, achievable joint velocity or acceleration, and dynamic motion capability of today’s humanoid robots are far more limited than those of the average human subject. Our research group focuses on motion control especially using robot to preserve Japanese traditional dances, which are considered as intangible cultural assets. In this work the upper body motions of robot is generated. First the original human motion is acquired by motion capture system. Then inverse kinematics is used to convert data in the form of markers’ positions into joint angle of robot. As the robot has many physical constraints, converted data often cannot be used directly, so the optimization process is used. The optimization algorithm’s objective function is subjected to preserve salient characteristic of the original motion, while constraints for optimization are used to transform the motion of human to the capabilities of the humanoid robot. Hierarchical B-spline is used as data representation for trajectory optimization. The B-spline function is exploited in all constraints for optimization, which makes our work different from previous trajectory optimization based methods that only reduce robot’s physical parameters such as velocity or force, but not effectively limit them. Generated motions are first simulated in the program developed by our research group, then the humanoid robot HRP2 is used to test optimized motions.