Unscented optimization and control allocation

• Main Authors: Karpenko, Mark, Ross, Isaac Michael, Proulx, Ronald J
• Format: Patent
• Language: English
• Published: 04.09.2018
• Subjects: CHAMBERS PROVIDED WITH MANIPULATION DEVICES; CONTROL OR REGULATING SYSTEMS IN GENERAL; CONTROLLING; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; HAND TOOLS; MANIPULATORS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS ORELEMENTS; PERFORMING OPERATIONS; PHYSICS; PORTABLE POWER-DRIVEN TOOLS; REGULATING; TRANSPORTING

The disclosure provides a robotic arm controller which determines a control parameter for at least one actuator comprising the robotic arm using a control equation having the general form Ax=b, where A is a transformation matrix A based on the geometry and Jacobian of the robotic arm, x is the control parameter x such as a torque vector at a specific joint, and b is the end effector parameter b which specifies a desired corrective state of the end effector. The methodology, by way of constructing and solving an unscented optimization problem, provides a solution to the Ax=b problem by perturbing at least one joint angle appearing in the Jacobian to generate a plurality of distributed joint angles, determining a control parameter x which minimizes an error function. In a particular embodiment, the error function is sum of residual squares, and the appropriate control parameter x is determined by minimizing the error function subject to a series of constraints Aix−b−zi=0, where each constraint arises by virtue of the error generated through use of a given joint angle in the plurality of distributed joint angles selected.