Giving Santos™ Dynamic Motion: Dynamic Motion Prediction
Autonomous behavior is predicted where initial and final configurations of the human are specified and the code predicts the natural motion in between. Optimization code is used to determine motions that are governed by the laws of physics as well as various human performance measures, and constrained by the restrictions imposed by the skeleton, physical constraints, and environmental constraints. By using optimization the equations of motion do not have to be integrated which makes the code fast and powerful and provides a very effective way for doing dynamics. Currently the human performance measures include minimizing total body energy, minimizing torque at the joints, and maximizing stability. The values of torque produced at each joint can be used to calculate muscle deformation and stress. Dynamic motion prediction is extremely important for Santos™ because it enables him to perform autonomous tasks like walking and carrying objects. These motions depend on the task as well as the magnitude of forces involved. The code is specialized to deal with specifi c tasks, this specialization increases efficiency and speed in an effort to achieve real-time simulations. Currently the two task-based programs include code for gait prediction and code for upper body motion prediction. A third programfor predicting climbing is being developed. The overall methodology for using the mature field of optimization has allowed for solving a complicated dynamics problem including the human body and can be extended to other fields.
