Control of pneumatic systems for free space and interaction tasks with system and environmental uncertainties

Abstract

This dissertation presents three control methodologies for pneumatic applications in both free space and constrained environment with system and environmental uncertainties. First, a model reference adaptive controller (MRAC) is designed to achieve accurate free space position tracking of a pneumatic actuator by estimating and compensating for friction uncertainties. Three parameters of a static friction model are estimated. The controller consists of an inner and outer loop structure. The inner loop provides the desired actuation force through adaptive estimation and the outer loop achieves the desired actuation force through a sliding mode force controller. Both loops are based on Lyapunov stability. Experimental results verify the stability of the controller and show that adaptive control improves position tracking accuracy and reduces payload sensitivity without tuning the friction compensation manually. Then, a passivity-based approach is taken to carry out stable and dissipative contact tasks with an arbitrary and unknown passive environment (unknown in terms of stiffness and location). A pseudo-bond graph model is developed to prove the passivity of a pneumatic actuator controlled by proportional valves. Using this model, an open-loop pneumatic actuator can be proven to not be passive, but it can be passified under a simple closed-loop feedback control law. The passivity of the closed-loop system is verified in impact and contact force control experiments. Finally, an energetically derived control methodology is presented to specify and regulate the oscillatory motion of a pneumatic hopping robot, which constantly switches between free space and a constrained environment (system and environmental uncertainties). The desired full hopping period and the desired flight time are predefined to solve for the static pressure in the upper chamber and the velocity immediately before lift-off. Therefore, during contact, the pressure in the upper chamber is controlled according to a position-based mapping to control the duration of contact, while controlling the total conservative energy of the system specifies the flight time. During flight, both chambers are sealed to preserve the passive dynamics of the system. This control methodology is demonstrated through simulation and experimental results to provide accurate and repeatable energetically efficient hopping motion.