| The wealth of behaviors exhibited in natural human motion results from the complex interplay of biomechanical and neurological factors. More specifically, the biomechanical factors include musculoskeletal kinematics and dynamics while the neurological factors include sensorimotor inte gration, planning, and control. Much of the difficulty in understanding and synthesizing human motion is related to the proper characterization of these factors, as well as the proper implementation of a system-level framework that models their interrelationships. Thus, the choice of an appropriate conceptual framework for human motion synthesis is critical and should be based upon a general and extensible motion control abstraction, which can be integrated with biomechanical and neurological models.; This thesis addresses motion control with application to human motion synthesis. A task/posture abstraction is used as a system-level framework for modeling human motion. In this framework motion is represented using a task space description complemented by a postural description. This framework also incorporates constrained motion strategies and computational muscle models. As such, it represents a coherent methodology for the management of motion tasks, physical constraints, and neuromuscular criteria. Some key contributions include the development of a task-level control methodology for constrained systems. The task/constraint partitioning approach contained within this methodology is applied to the control of simulated human and robotic shoulder complexes. Addressing the postural component of human motion, a muscle effort criterion is developed and applied to the prediction of upper limb postures. Finally, the design and implementation of a musculoskeletal simulation framework is presented and some areas for future work are discussed. |
