Contents I. Introduction
A Mobile Manipulator (MM) system consists of a standard robotic arm equipped with a movable platform. Such MM is mostly used in many places such as deactivation of explosives, load transportation, exploration, and dangerous environments maintenance. Many researches on MM systems have aroused due to these primary applications. Such a MM system incorporates the advantages of large workspace capabilities of the mobile platform along with increased manipulator dexterity and hence such robots are appropriate for those purposes which require both manipulation capabilities and motion. The modeling is complex due to the coupled dynamics of MM and mobile platform non- holonomic constraints, irrespective of these advantages. Even though the combined system is complex, when subsystems are considered separately, it takes advantage of robotic arms and mobile platforms and reduces their drawbacks. Nowadays, to achieve good tracking capabilities, many effective control schemes have been developed for MM systems. There are two approaches for the control of such systems. In first approach, the whole MM system is considered as two subsystems which are interconnected: the manipulator arm subsystem and the mobile platform subsystem. In this category, decentralized control techniques are used. The controllers for these two subsystems are created separately and communications between the platform and the manipulator can be either considered or neglected. In the second approach, the MM is viewed as a single system in which a single controller is developed for the whole system. Many methodologies have been developed for MM considering it as a single system. In [1], an extended Jacobian approach for reducing the arm vibrations of a MM is discussed. In [2], trajectory tracking of a MM based on decentralized control in the Cartesian space is discussed. In [3], trajectory planning based on point-to-point control for a MM is done with a differential integrated control method. A planning control scheme for the coordination of the motion of a MM is presented in [4]. A control strategy is presented in which the manipulator is positioned at the preferred configuration based on the measurement of its manipulability. For a MM which is operating in the task space, the control problem for position control with state constraints is proposed in [5]. For the whole MM system a Proportional Derivative (PD) along with neural network controllers is proposed in [6]. In [7], an adaptive controller is proposed which ensures that even in the presence of system coupling, the state of the MM system could follow the required trajectories.
