Abstract by Gabriella Smith
Analyzing Linear Approximations of Robotics Systems.
Engineers at the BYU Robotics and Dynamics Laboratory(RAD LAB) have been experimenting with optimal control algorithms to control new robot designs. Due to the complexity of these robotics systems the computational approach that has been developed is to use a linear approximation of the robots dynamics. The two algorithms that were developed for this purpose are called (i) fixed state and (ii) coupling torque. In the fixed state algorithm, we assume a fixed coupled linear system between each time step. In the coupling torque algorithm, we approximate the coupled terms of this linear system with a constant torque. In order to understand the robustness and utility of these algorithms we have numerically explored a variety of parameters associated with these methods. Specifically, we are experimenting with the perturbations to our system’s dynamics and analyzing their effects on both the fixed state and coupling torque algorithms.