Sequential Re-planning for Dextrous Grasping Under Object-pose Uncertainty

This work shows how successive grasp attempts can be re-planned to make use of tactile information acquired during previous grasp attempts. Our main contributions are to enable planning of dexterous grasping for high degree of freedom manipulators, and belief updating from tactile sensors in 6 dimensional space. The method is demonstrated in trials with simulated robots. Sequential re-planning is shown to achieve a greater success rate than single grasp attempts, and trajectories that maximise information gain require less replanning iterations than conventional trajectories before a grasp is achieved.