A behavior-based control system for mobile manipulation

Abstract: The field of industrial robotics can be defined as the study, design and use of robot manipulators for manufacturing. Although the problem of designing a controller for industrial robots has been subject of intensive study, a number of assumptions are usually made which may seriously limit the applicability of these robots. First, the robotic manipulator is usually considered to be positioned at one place, which means that it can only work in its limited working envelope fixed to this position. Second, it is usually assumed that the environment of the manipulator (workcell) is carefully engineered to suit the task and the configuration of the arm. Finally, the control program of the manipulator is often designed assuming that the task will not change. These restriction make current industrial robots unsuitable for the new demands of flexible automation in small and medium enterprises. In this thesis, we develop techniques that extend the applicability of current robotic manipulators, by addressing theabove limitations. We propose an approach to sensor-based manipulation that: 1) has flexible and modular control system, in order to easily to new tasks and environments, 2) the execution is sensor-based for robustness in less controlled environment, and 3) our approach applies to the more general problem of combined mobility and manipulation, in order to extend the work space of the manipulator.