Optimization of parallel robot structure using screw theory

Alon Wolf

Abstract

My talk will be divided into two parts the first will deal with the optimization of the kinematic parameters, of a parallel robot structure, for a given task. Drawing principally upon tools from screw theory, a parametric objective function which describes the robots' performance, with respect to the given task, is derived. This function represents the sum of the reciprocal product of a twist, which defines the task, and the wrenches described by the governing lines of the robots' structure. Physically, the objective function quantifies the instantaneous work generated by the robotic structure to its platform while the latter undergoes an instantaneous motion in the twist (task) direction. The parameters which are being optimized can represent the location and geometry of the robot. This method can be used in the designed stage of a task-oriented new robot or while performing tasks using a fixed geometry robot or a reconfigurable robotic structure, capable of changing its geometry. Several examples are provided including an optimal design of a mini parallel robot for spinal operation, and optimization of the 3-UPU parallel robot.

The second part of the talk will deal with geometric interpretations of the derivatives of the Jacobian matrix of parallel robots with application to stiffness control.

 

Speaker Appointments

For appointments, please contact Alon Wolf (alonw@andrew.cmu.edu).