Abstract
A distributed control method for hyper-redundant manipulators is proposed, aiming at applying to complicated and unknown or dynamic environment. A manipulator consists of serially connected joint units, and each unit has the same degrees of freedom as dimension of task space. The proposed method is highly distributed: each unit is controlled by one controller, only neighboring units communicate with each other, and each unit does not have to obtain information from far units. The method is multiple point control: not only the end-effector but also several units can converge to their own desired positions stably. Thus the manipulator can avoid obstacles only if the units which detect the obstacles with their own proximity sensors set their desired positions far from the obstacles. The manipulator makes expansion and contraction motion; it can move into/out of long and narrow space like pipes. All units are controlled so that their displacements may be equal. Hence they have the same and maximum displacement margin. The dynamics of the end-effector is not influenced by the motion of units; that is important for doing tasks with the end-effector. The proposed method is applied to a planar hyper-redundant manipulator with rotational joints, and the effectiveness and usefulness of the method are ascertained by computer simulations and experiments.
