Fig. 6

More details about the regrasp graph. The regrasp graph has three layers where the top layer and the bottom layer have one circle , respectively. They encode the initial and goal states and their accessible grasps. The correspondence has been shown by the red arrows in Fig. 5. The middle layer encodes the placements on horizontal surfaces and their accessible grasps. The four placements and their accessible grasps of Object B are plotted in the figure. Their correspondent circles in the middle layer are denoted using red arrows. By searching this regrasp graph, the planner finds a sequence of grasps. The sequence may include some intermediate nodes like the one shown in the lower part of Fig. 5. It may also connect directly from the initial state to the goal