Hand Control
Several Controllers have been tested on DLR Hand II. At the level of individual fingers, Cartesian and joint-level impedance controllers have been implemented. Also force control has been tested. At the level of coordinated control for multiple fingers, a coordinated impedance control on the impedance of the object has been applied.
Detection of the Contact Points on the Finger Tips
When grasping and manipulating an object by the tips of the fingers, the location of the contact on the surface of the fingers is essential to determine manipulation trajectories as well as to compute the appropriate grasping force. At the Institute of Robotics and Mechatronics, an algorithm has been developed, that allows to estimate this contact position even when no tactile sensors are present. The algorithm borrows from the observation that when multiple fingers grasp an object, the mobility of the object for in hand manipulation is limited such that the contacts between the fingers and the object does not break. This limited motion can be used to estimate and observe the position of contact directly or during ongoing manipulation in an information filter similar to Kalman filtering.
Detection of the Contact Point on the Object and Localization of the Object
On the other hand, when a model of a grasped object is available, for example from the measurements of a laser range scanner, the location of the contact points on the surface is of interest for the planning of particular tasks as insertion of the object into a box or similar. A algorithm has been developed to compare the measurements or estimates of contact location at the finger tips and the surface normal at the point of contact to an object model. Hypothesis for possible contact locations on the object can be developed and tested for their probability. In this fashion the most likely contact positions can be determined. This can also be evaluated to determine the position and orientation of the object within the grasping hand.