Bimanual In-hand Manipulation using Dual Limit Surfaces

TL;DR

This paper introduces a modeling and planning framework for in-hand object reconfiguration, focusing on frictional patch contacts between the robot's palms (or fingers) and the object, which leverages two cooperative patch contacts on either side of the object to iteratively reposition it within the robot's grasp by alternating between sliding and sticking motions.

Abstract

In-hand object manipulation is an important capability for dexterous manipulation. In this paper, we introduce a modeling and planning framework for in-hand object reconfiguration, focusing on frictional patch contacts between the robot's palms (or fingers) and the object. Our approach leverages two cooperative patch contacts on either side of the object to iteratively reposition it within the robot's grasp by alternating between sliding and sticking motions. Unlike previous methods that rely on single-point contacts or restrictive assumptions on contact dynamics, our framework models the complex interaction of dual frictional patches, allowing for greater control over object motion. We develop a planning algorithm that computes feasible motions to reorient and re-grasp objects without causing unintended slippage. We demonstrate the effectiveness of our approach in simulation and real-world experiments, showing significant improvements in object stability and pose accuracy across various object geometries.

Figures (6)

• Figure 1: In-hand manipulation of a grasped object using frictional patch contacts. The object maintains sticking contact with the lower palm, while maintaining sliding contact with the upper palm. As the lower palm moves, the object's position changes with respect to the upper palm, which enables object regrasping.
• Figure 2: Iterative repositioning of the grasped object by alternating sticking and sliding contact between the two robot palms.
• Figure 3: Visualization of the wrench space for bimanual manipulation. The red ellipsoid represents $\bm{w}_a\bold{A}\bm{w}_a = 1$ while $\bm{w}_b\bold{B}\bm{w}_b = 1$ is the blue ellipsoid. The green region shows the slippage-free twist constraint. A non-convex slippage-free green region occurs when the blue ellipsoid covers over half of the red ellipsoid surface, shown in [a]. A convex slippage-free green region occurs when the blue ellipsoid covers less than half of the red ellipsoid surface, [b].
• Figure 4: Simulated bimanual manipulation setup using Franka Emika Panda Robot Arms in Drake. We model patch contacts with the object using hydroelastic contacts, shown in red and green.
• Figure 5: Real-world experimental setup: 7 DOF robot arm with a 6 DOF force/torque sensor, a transparent palm end-effector, and a camera in the palm to detect object pose, shown in [a]. Two object geometries -- square, circle -- each made of brass with self-engaging tape on the surface, [b].