A computational model of prehensility and its application to dextrous manipulation

The authors formulate the model of computation in two parts: an analysis part in which a grasp-based model is represented as a collection of topological and geometric spaces described at various levels of detail, and a synthesis part in which the required transformations to achieve dextrous manipulation are described as a combination of two classes of processes: a postural transformation process (to derive an appropriate hand posture); and a functional transformation process, (to derive an intended hand functionality). It is shown that both the hand posture and the hand functionality can be computed from the symbolic high-level task requirements and object attributes, and can be transformed into numeric low-level joint space variables. Partial simulation results from applications to dextrous manipulation and discussions are given