In recent years, there has been considerable research performed in the area of mobile manipulators subject to kinematic constraints. This thesis addresses the position control problem of mobile manipulators subject to kinematic constraints. The main objective is to develop practical controllers that can be easily implemented on a mobile manipulator in the presence of parameter uncertainty and unmodeled disturbances, such as terrain irregularity and unknown payload.

In this thesis, two mobile manipulator controllers are developed:​ (i) a neural network-based hierarchical intelligent controller;​ and (ii) a hierarchical robust controller, each aiming at different control purposes. The proposed hierarchical intelligent controller is based on 'radial basis functions'. As such it has very good learning ability and adaptability, thus it is especially suitable for applications where mobile robots move in unknown and changing environments. When the adaptability is less demanding ('e.g.', the environment is partially or completely known), or when the available system resources, such as size of embedded memory and on-board computational capability are restricted, the hierarchical robust controller is suggested as an alternative. It has a simple control structure and a low computational load.

Both proposed controllers require the measurement of velocity signals. In practice, velocity sensors may have to be omitted due to considerations of size and weight. In this case the velocity signals are determined by a first-order numerical differentiation of available position signals. This may introduce significant noise in the control signal and degrade its performance. To resolve this problem, in this thesis we propose a new reduced-order adaptive observer-based velocity measurement technique requiring only position measurements.

Rigorous stability and performance analysis of the developed theory is also provided in the thesis. In addition, extensive experiments were conducted on a four-degree-of-freedom robotic manipulator. The experimental results validate the effectiveness of the proposed controllers.