An Operational Space Tracking Algorithm for Generating Dynamic Simulations of Movement

The objective of the current project was to develop a computationally practical method for generating forward dynamic simulations of three-dimensional movements that track experimental data while allowing for bias errors in joint angle measurements. An operational space control framework was adapted from robotics and used to drive a 20 degree-of-freedom, torque-actuated simulation of normal gait. Joint torques were computed that precisely tracked the trajectories of the feet (end-effectors) while also generally tracking the angular displacements of the lower-extremity joints. The algorithm generated a forward dynamic simulation of gait using about 30 minutes of computer time. During stance, adjustments to the experimental joint angles were introduced that removed penetration of the feet into the ground and sliding of the feet relative to the ground. Simulated joint torques were similar in both shape and magnitude to what has been computed using inverse dynamics. We believe that an end-effector control formulation provides a powerful simulation framework in biomechanics in which movement can be described with greater emphasis on the task rather than on traditional joint-space representations