In this thesis, we tackle the problem of robust simultaneous position and force control of cooperative robotic (CR) structures in the presence of ununmodel known time-varying structured and unstructured uncertainties. Most non-adaptive control schemes for the coordinated control of multiple manipulator systems, usually assume a full knowledge of the system's dynamics. This is an unrealistic assumption in many cases since these complex systems are usually subject to the ubiquitous presence of uncertainties. If not dealt with appropriately, these uncertainties may have a dramatic effect on the controller's performance and may even induce instability. Although recent research work carried out in the area of conventional adaptive control has led to a significant improvement in the tracking performance of, both, the payload's desired position/orientation and the internal force desired values, in the face of parametric uncertainties, the majority of it has ignored the effect of unstructured uncertainties on the controller's performance and its stability. Modeling imperfection of complex systems, such as closed-chain robotic manipulators, is inevitable. This makes the development of a robust control approach for the increasingly complex cooperative manipulator systems a necessary step to keep up with the increasingly demanding design requirements of such systems.

In this thesis, we develop novel approaches based on soft computing tools to tackle this complex, yet important control problem. A special type of adaptive fuzzy controllers is first designed to learn the system's overall dynamics without a prior knowledge of it. The controller is then improved even further to provide for a more efficient behavior particularly with respect to computational complexity. Both soft computing based controllers are shown to have excellent tracking abilities of the payload's desired position/orientation while meeting the internal force desired values. The controllers are also shown to be highly robust in the face of a substantial amount of parametric and modeling uncertainties with varying intensity levels. It is also formally proven that, under a few reasonable assumptions, the position and the internal force tracking errors always converge to zero. The results obtained in this work have been very encouraging and would certainly open new opportunities for tackling robot control of complex structures in general. This being said, we believe some more improvements could be done to make the approaches even more powerful and readily implementable in real world environments.