| This dissertation presents a two-stage six-dimension wrench space grasp synthesis method by using model-based simulation and computational geometry. The first phase is to obtain sets of finger contact points on the graspable portion of a given object using a geometrically accurate offline simulation of a desired gripper. The contacts are considered as unit-magnitude point contacts with friction, and a pyramidal friction cone approximation is used. The six-dimension convex hull of the generating wrenches for each grasp is computed, and all grasp information is then archived. In the second phase, which is our main contribution, the weight of the object (or any task-related external force) is used to compute an external disturbance wrench, which is considered to be a point in six dimensional wrench space. The archived grasps are then searched for the one that is closest to the disturbance wrench using our six-dimension computational geometry algorithm. This grasp will have the minimum peak force to achieve equilibrium. An example of the method is shown. |
