| Our goal is to design a flexible, scalable system of multiple robots that can pick up, transport, and lower different types of objects with minimal human intervention. Each robot consists of a computer controlled mobile platform with a manipulator. The system must be robust to the robots' positioning errors and must be able to grasp the object stably. This dissertation addresses the fundamental issues that are relevant to the design, control, and analysis of such a system.; As the robot grasps the object, the forces must be controlled so that the object neither slips nor is crushed. We designed an arm specifically for the task of grasping objects without using special purpose fixtures. The fixtureless approach is flexible and our design allows the forces to be easily controlled. The arm enables multiple robots to cooperate because it can accommodate robot positioning errors as well as control the grasping forces.; Our approach to the coordination of robots allows the use of heterogeneous mobile platforms. The platforms can be omnidirectional or nonholonomic. Further, they may be equipped with passive or actively controlled arms. Our contributions to robot control are twofold. First, the individual mobile platforms must be controlled precisely before they can be used in a multi-robot system. We designed a controller that uses state feedback that will exponentially converge the robot to the correct position and orientation. Second, we developed controllers for robust manipulators that control the force interactions between the different robots. Specifically, we developed a novel active stiffness controller that allows the locomotion and manipulation problem to be decoupled mechanically, thus allowing for simpler control.; Lastly, the individual robots must be placed around the object and the individual contact stiffness must be designed. We focused on two measures, the first being grasp stability which measures the ability of the system to reject wrench disturbances. The second measure focused on the effects of the robots' positioning errors on the individual robot grasp forces, and the ability to be robust to these errors. The new analysis can easily be applied to a system of multiple robots.; Finally, we developed an experimental system of heterogeneous robots with active or passive arms mounted on each mobile platform. Our experimental results show that such a system is capable of picking up and transporting a wide variety of objects without special purpose fixtures. |
