Depth-first iterative-deepening: an optimal admissible tree search
Artificial Intelligence
Planning Algorithms
| Hi-index | 0.00 |
In previous works a novel flexible and versatile handling concept, called PARAGRIP(Parallel Gripping), was introduced. This concept is based ona reconfigurable architecture with a modular and alterable layout. The robot system is able to handle objects with six DOF by forming a parallel kinematic structure including several robotic arms and the object itself. As many kinematic parameters, like the grasp- and base-points of the arms as well as the arm combination can be designed freely, the handling system offers a fast and economic possibility to adapt the performances to the requirements of the task. This adaption can proceed before or even during manipulation. The latter is realized by dynamic re-grasping, where the object is passed from one arm to the next, if more than three arms are available in the layout. This Paper deals with the question how an optimal configuration set can be planned automatically, if the robot layout offers the possibility of dynamic re-grasping. It shows the benefits and the challenges as well as the strategies of the planning process and the realization. The focus of this paper is how to manage the complexity of this numerousness of configuration possibilities and choose the optimal one within the shortest computation time.