Improving regrasp algorithms to analyze the utility of work surfaces in a workcell

Abstract

The goal of this paper is to develop a regrasp planning algorithm general enough to perform statistical analysis with thousands of experiments and arbitrary mesh models. We focus on pick-and-place regrasp which reorients an object from one placement to another by using a sequence of pickups and place-downs. We improve the pick-and-place regrasp approach developed in 1990s and analyze its performance in robotic assembly with different work surfaces in the workcell. Our algorithm will automatically compute the stable placements of an object, find several force-closure grasps, generate a graph of regrasp actions, and search for regrasp sequences. We demonstrate the advantages of our algorithm with various mesh models and use the algorithm to evaluate the completeness, the cost and the length of regrasp sequences with different mesh models and different assembly tasks in the presence of different work surfaces. Our results show that spare work surfaces are beneficial to assembly. Tilted work surfaces are only sometimes beneficial, depending on the objects.

A Certified-Complete Bimanual Manipulation Planner

“…However, the authors of [32] also proposed an interesting notion of Grasp-Placement Table, based on the discretization of G ∩ P, which captured the connectivity of G ∩ P. More recent work on regrasping such as [5], [6] also employed some kinds of graphs to represent the connectivity.…”

Analyzing the utility of a support pin in sequential robotic manipulation

“…Jiang [9] outlined a learning-based framework to determine how to place novel objects in complex placement areas, while Baumgartl et al [10] computed the geometric information of both the placement area and the object to determine whether a location is suitable for placement. Once the placement location is determined, the stable pose of the object at that location is computed by applying the contact constraint and the stability constraint [6], [10], [2]. Other constraints such as an orientation preference for man-made objects [11], [9] can also be taken into account by using the learning-based framework.…”

“…It is desirable if the object has many different ways of placements and each placement has many valid grasps, because this can provide the robot with more choices to incrementally adjust the object's pose. More formally, the flexibility in placements helps to increase the connectivity of the regrasp graph [2] and is crucial for the quality of the resulting regrasp sequence.…”

“…The main advantage of a support pin lies in its ability to greatly increase the number of stable placements as well as the directions from which an object can be manipulated. Our previous work studying the use of a tiled surface for regrasping ( [2], [5]) demonstrated the benefits of these two properties on the success rate of sequential robotic manipulation. Since a pin is able to provide more intermediate states and more candidate regrasps than those through usage of a tilted surface, we expect it to have better performance while planning the regrasping sequence.…”

“…Our algorithm automatically computes the stable placements of an object on a support pin, finds force-closure grasps, generates a graph of regrasp actions, and searches for regrasp sequences. We use the two-layer regrasp graph in [2] to decouple the search of pick-and-place sequence and the search of grasps, and delay expensive inverse kinematics and collision detection computations until necessary. We evaluate the success rates of the tasks and the length of regrasp sequences with different mesh models, different pin lengths, and different tasks including reorientation (i.e., flipping) and assembly.…”