The developed mobile manipulator is primarily composed of a mobile base, a robot manipulator and an eye-in-hand vision system. The material handling of a mobile manipulator has two stages: guiding the mobile base between stations, and picking up a workpiece from a station. Fast landmark recognition and obstacle detection based on color segmentation are proposed for path following, obstacle avoidance and mobile base positioning. Using the machine vision, a vision-based vector field histogram method is modified and applied to guide the mobile manipulator for obstacle avoidance. However, after the mobile manipulator arrives at the station, positioning errors of the mobile base and the non-horizontality of ground inevitably cause position and orientation errors of the mobile base relative to the station. A vision-guided control strategy with a behavior-based look-and-move structure is presented. This strategy is based on six predefined image features. In the designed neural fuzzy controllers, each image feature is taken to generate intuitively one degree of freedom motion command relative to the camera coordinate frame using fuzzy rules, which define a specific visual behavior. These behaviors are then combined and executed in turns to perform grasping tasks.

This study constructs a wheeled humanoid robot as our preliminary research of humanoid robots. The humanoid robot developed herein comprises mainly a wheeled mobile base, a fixed torso mounted on a mobile base, two 7 DOF robot arms, two 7 DOF robot hands and one 5 DOF robotic binocular head. A coordinated visually guided control structure is then proposed for this humanoid robot. Image processing algorithms are also proposed to increase the perception capability of rapidly detecting the location of a target object in a natural environment, thus reducing the computational burden for image processing. Finally, one experiment is conducted to verify the theoretical derivations and the performance of the humanoid robot. Three objects, i.e., a rectangular parallelepiped, a cup with a handle and a bottle-shaded container with square bottom, are placed on a table, with one selected as the target object. Experimental results indicate that the robotic head and the robot arm of the humanoid robot can coordinate with each other to locate and grasp the target object successfully.

Mobile manipulators, which consist of a mobile base and a robot manipulator equipped with a vision system, are appropriate for transferring small quantities of a range of different materials in production lines. Position and orientation errors of the mobile base relative to the station are inevitably caused by the non-horizontality of the ground and positioning errors of the mobile base. Hence, this study utilises an uncalibrated eye-in-hand vision system to provide visual information for controlling the manipulator mounted on the mobile base, to pick up a workpiece located on the station. A vision-guided control strategy is proposed. It is based on selected image features and an image-based look-and-move control structure. The control law employed in the control structure is based mainly on the off-line estimate of the transformation from feature space to Cartesian space using the least squares estimation algorithm. Finally, the positioning performance of the eye-in-hand manipulator is experimentally evaluated by controlling the end-effector of the manipulator to approach and grasp the workpiece in various locations on a station.

Mobile robots frequently replace humans in handling and transporting wafer carriers in semiconductor production lines. A mobile robot is constructed in this paper. The developed mobile robot is primarily composed of a mobile base, a robot manipulator, and a vision system. Since the guidance control system of the mobile base inevitably causes positioning errors of the mobile base, this study employs the eye-in-hand vision system to provide visual information for controlling the manipulator of the mobile robot to grasp accurately stationary material during pick-and-place operations between a predefined station and the mobile robot. This work further proposes a position-based look-and-move task encoding control strategy for eye-in-hand vision architecture, that maintains all target features in the camera's field of view throughout the visual guiding. Moreover, the manipulator can quickly approach the material and precisely position the end-effector in the desired pose. Numerous techniques are required for implementing such a task, including image enhancement, edge detection, corner and centroid detection, camera model calibration method, robotic hand/eye calibration method, using a camera with controlled zoom and focus, and task encoding scheme. Finally, the theoretical results for the proposed control strategy are experimentally verified on the constructed mobile robot. Specific experimental demonstrations include grasping the target object with different locations on the station and grasping the target object tilted by different angles to the station.

A new nonlinear manipulator control scheme is proposed in this paper which is robust against modeling errors and unknown disturbances. It is based on the principle of sliding mode control (SMC) but with the application of fuzzy logic control (FLC). The control method provides a simple way to achieve asymptotic stability of the system. In addition, the method can not only overcome the chattering problem caused by sliding mode control, but also determine the rules and parameters for fuzzy logic control. Experimental results are presented to verify the theoretical derivations.>

The control strategy of two cooperating robots is described. A two-fingered hand is an example of such a system. A control architecture is proposed that guarantees: (1) each robot end-point follows its position input command vector 'closely' when the robots are not in contact with each other, and (2) when the two robots come in contact with each other, one robot controls the position of the contact point, while the other controls the contact force. The controller does not need any hardware or software switch for the transition between unconstrained and constrained maneuvering of each robot.>

A nonlinear approach for the stability analysis of robot manipulators in constrained maneuvers is presented. Stability of the environment and the manipulator taken as a whole has been investigated, and a bound for stable manipulation has been derived. The authors show that for stability of the robot, there must be some initial compliancy either in the robot or in the environment. The general stability condition has been extended to the particular case where the environment is very rigid in comparison with the robot stiffness. The stability analysis has been investigated using unstructured models for the dynamic behaviour of the robot manipulator and the environment. This unified approach of modeling robot dynamics is expressed in terms of sensitivity functions as opposed to the Lagrangian behavior of all the elements of a robot manipulator (i.e., actuators, sensors, and the structural compliance of the links) in addition to the rigid body dynamics.>