UBC Theses and Dissertations
Observer based velocity and environment force estimation for rigid body control Hacksel, Peter J.
Issues pertaining to the control of a rigid body are considered in this work. A magnetically levitated robotic wrist is here modelled as a rigid body subjected to both control and environment forces. All the algorithms presented here were implemented on the Maglev wrist for experimental verification. A good inertial model of the rigid body was needed for the estimation and control work in this work; a least squares identification procedure, found in the literature, was implemented for this purpose. Algorithms are developed for obtaining dynamic environment force and torque estimates from position errors in a velocity observer. Obtaining torque estimates from a nonlinear velocity observer is achieved by using the linear force observer as an analogue. Stability of the torque observer to a constant environment torque is proven for the one dimensional case only. Simulation and experiments are used to verify the stability in the three dimensional case. Environment force/torque measurements, alone useful for control purposes, are also used to correct a velocity observer. The resulting corrected observer design gives low noise velocity estimates and is robust in that it is not sensitive to errors caused by interaction with the environment. As a demonstration of the merit of the observer based work, a remote center of compliance controller is implemented. An inertial model is needed to provide feedforward that linearizes the dynamics to a. second order system. Corrected observer velocities are used in the feedforward and the derivative component of the control law. The linear dynamics about a remote center is verified by comparing the environment forces measured with the velocities observed.
Item Citations and Data