Feedback linearization is an effective controller-design methodology for nonlinear systems where it is difficult to obtain a finite number of operating points to linearize the system for designing well-known linear robust controllers. Feedback linearization becomes one of very limited methodologies that can be used for control of such systems. Traditional implementations of feedback linearization technique are not robust, which means this control methodology does not account for system uncertainties. The reason being that the control law methodology assumes accurate knowledge of nonlinear dynamics of the system. Recently, in [1] a new methodology was proposed which adds robustness to feedback linearization. The methodology uses sensitivity dynamics-based control synthesis. The methodology was demonstrated on a simple proof-of-concept single actuator mass-spring-damper model.

This research is focused on application of robust feedback linearization technique to real life complex hydraulically actuated physical systems. In particular, the methodology is applied to the problem of controlling mechanical linkage configuration in excavator machines. The problem addressed is controlling of bucket angle of excavator such that the bucket is always kept parallel to ground irrespective of boom motion to avoid spilling of the load. The dynamics of systems such as excavator linkage actuated by hydraulic actuator are often complex and application of robust feedback linearization (RFL) methodology gets tedious and cumbersome. The work in this thesis is intended for demonstrating the applicability of RFL methodology for such complex systems and also to lay foundation for development of an automated user-friendly toolbox to enable easy use of such control technique in day-to-day practice. The uncertain parameter considered in the development in this thesis is the bulk modulus of the system as it is the most common uncertainty in the system. The modeling process also considers portability of models from some known commercial software tools such as SimHydraulics and SimMechanics.

The results presented show that the RFL methodology is very effective in achieving robust control of hydraulically actuated systems with uncertainties in hydraulic parameters.