International Journal of Humanoid Robotics

The development of the flutist robot at Waseda University since 1990 has enabled a better understanding of the motor control functions required for playing the flute. Moreover, it has introduced novel ways of interaction between human beings and humanoid robots such as: performing a musical score together in real time and transferring skills to flutist beginners. In this paper, the development of the Waseda Flutist Robot No. 4 Refined (WF-4R) is presented. The mechanical design of the components of the robot and the control architecture are detailed. In order to efficiently control and coordinate the motion of each of the simulated organs of the robot, an algorithm was proposed to extract the features required to perform a score based on human performance. This algorithm was divided into two phases: sound calibration and music score performance. Finally, an experimental setup was done to verify the effectiveness of each of the phases by analyzing the time and frequency domain responses from recordings of the robot performances. The WF-4R is able to perform from musical scores quite similar to human.

One of basic characteristics of the regular bipedal walk of humanoid robots is the maintenance of their dynamic balance during the walk, whereby a decisive role is played by the unpowered degrees of freedom arising at the foot–ground contact. Hence, the role of the Zero-Moment Point (ZMP) as an indicator of dynamic balance is indispensable. This paper gives a detailed discussion of some basic theoretical assumptions related to the ZMP in the light of imprecise, and even incorrect, interpretations that have recently appeared, and which have led to some erroneous conclusions. Examples are given to show some erroneous basic attitudes and the genesis of some of them is indicated. It is also pointed out that in the domain of bipedal walk there are still notions that are not clearly defined and their meanings differentiated in some related branches of science and engineering. One of the examples is dynamic balance and stability, which are often used interchangeably.

For 3D Euclidean reconstruction, the challenging problem is that only the hypothesis of intrinsic parameters can be used to retrieve the camera parameters without additional information. In this paper, we propose a method to find out the intrinsic parameters of a camera using the rank constraint of the relation matrix of absolute conic Ω. The degenerative problems and the problem of variable internal parameters are also studied. The experimental results presented show the good performance compared with the results of other methods for the self-calibration of a camera.

We present a sophisticated Lagrangian model for anatomically and physiologically realistic human biodynamics (RHB), to accompany the recently reported Hamiltonian formulation.¹ The present RHB formulation is designed around three main modules: (i) A Riemannian configuration manifold, composed of gauge Lie groups of constrained 3D rotations and translations, which includes more than 300 degrees of freedom (DOF); (ii) exterior Lagrangian dynamics of the human musculo-skeletal system, including all natural conservative, dissipative and driving forces, powered by 600 equivalent muscles; and (iii) hierarchical nonlinear control, based on an iterative Lie derivative formalism, resembling both spinal reflexes and coordination-control of the human cerebellum.

RHB is driven by individual, user supplied musculo-skeletal data. It is modeled in the computer algebra system Mathematica™, simulated in Delphi™ and animated in the 3DS Max™ graphical environment. As an applied example of RHB, we present the full spine simulator, with 150 DOF (25 movable joints each with three constrained rotations and translations), muscular excitation and contraction dynamics, spring-and-damper ligament-like dynamics, spinal-like and cerebellar-like control, and external torques and forces (including inertial, gravitational, viscous, elastic and various types of impacts).

We propose a new human inspired structure of the lower extremity mechanism by which a humanoid robot will be able to efficiently perform fast movements such as running and jumping. We build a dynamic model of the humanoid robot which includes an elastic model of the biarticular muscle gastrocnemius and determine the role of the biarticular muscles and the elastic tendons in performing the vertical jump. We demonstrate that biarticular links contribute a great deal to the performance of the vertical jump. We also show that timing of the biarticular link activation and stiffness of the biarticular link influence the height of the jump considerably.