Biomechanical modeling and control of the human body for computer animation

Realistic anatomical modeling capable of high-fidelity synthesis of human body shape and motion is a major challenge in computer animation. Despite significant progress in this domain, the detailed modeling of the human body has not received adequate attention because of its complexity. We develop a comprehensive biomechanical model of the human body, confronting the combined challenge of modeling and controlling more or less all of the relevant articular bones and muscles, as well as simulating the physics-based deformations of the soft tissues. Emulating the relevant anatomy, our skeletal model comprises 75 bones (165 degrees of freedom), including the vertebrae and ribs, and it is actuated by 846 muscles, modeled as piecewise uniaxial Hill-type force actuators. To simulate the biomechanics of the soft tissues, we employ a coupled 3D finite element model with the appropriate constitutive behavior, in which are embedded the detailed anatomical arrangement and geometries of skin, muscle, and bone. As a precursor to developing our comprehensive biomechanical model, we consider the highly important head-neck-face complex. Our head-neck model is characterized by appropriate kinematic redundancy (7 vertebrae) and muscle actuator redundancy (72 muscles). It presents us with a challenging motor control problem, even for the deceptively simple task of balancing the head in gravity atop the cervical spine. Our biologically inspired, neuromuscular controller provides the numerous muscle actuators with efferent activation signals, controlling the pose of the head through time, as well as the stiffness of the neck by coactivating mutually opposed muscles. Using machine learning techniques, the neural networks within the controller are trained offline to efficiently generate the online control signals necessary to synthesize various autonomous movements for the behavioral animation of the human head and face. Our biomimetic modeling approach heightens the need to accurately model skeletal joints. Since the elementary joints conventionally used in physics simulation cannot produce the complex movement patterns of biological joints, we also introduce a new joint model, called “spline joints”, that can emulate biological joints more accurately. Spline joints can be efficiently simulated using minimal-coordinates-based dynamics algorithms.