Design, implementation, and testing of a robust adaptive control scheme for under-actuated non-linear systems

Abstract

Non-linear underactuated dynamic systems represent a very specific and challenging control problem. Their non-linear dynamics prevent them from being controlled with traditional linear techniques, and having some degree of underactuation further limits the existing control techniques which can be applied to these systems. This problem is even further complicated when a precise model of the system is unavailable. Such systems are widespread in the modern world, with real examples such as robotic manipulators and autonomous drones being implemented on ever greater scales. In order to address this, a combination of non-linear optimal and adaptive system identification and control techniques will be used. Simulations show that these techniques not only show promise in stabilizing these systems, but also in disturbance rejection, and, to a lesser extent, trajectory tracking. This is then validated by applying the proposed techniques to real world systems, thus proving that these techniques constitute an effective general hybrid optimal and adaptive approach to the identification and control of uncertain, underactuated, non-linear systems.