Posture optimization for pre-grasp interaction planning

Abstract

Many manipulation tasks involve grasping objects that are movable, not fixed, in the environment. An object’s potential for motion does contribute to the challenge of estimating its pose with sufficient certainty for robust grasping. However, object movability also offers an opportunity for pregrasp interaction strategies that adjust an object’s placement in order to improve grasping conditions. Here we highlight the results of initial work that illustrate the potential utility of pregrasp interaction for object acquisition into a desired grasp. We also present recent developments for refining the manipulator posture for pre-grasp rotation with respect to a payload cost metric. Optimization of the payload metric increase the safety margin with respect to uncertainty in the estimate of object weight.