Table of Contents
Fetching ...

Robust In-Hand Manipulation with Extrinsic Contacts

Boyuan Liang, Kei Ota, Masayoshi Tomizuka, Devesh Jha

TL;DR

A robust planning method is proposed that refines the motion cone to maintain desired contact mode regardless of parametric errors in the presence of kinematics parameter errors.

Abstract

We present in-hand manipulation tasks where a robot moves an object in grasp, maintains its external contact mode with the environment, and adjusts its in-hand pose simultaneously. The proposed manipulation task leads to complex contact interactions which can be very susceptible to uncertainties in kinematic and physical parameters. Therefore, we propose a robust in-hand manipulation method, which consists of two parts. First, an in-gripper mechanics model that computes a naïve motion cone assuming all parameters are precise. Then, a robust planning method refines the motion cone to maintain desired contact mode regardless of parametric errors. Real-world experiments were conducted to illustrate the accuracy of the mechanics model and the effectiveness of the robust planning framework in the presence of kinematics parameter errors.

Robust In-Hand Manipulation with Extrinsic Contacts

TL;DR

A robust planning method is proposed that refines the motion cone to maintain desired contact mode regardless of parametric errors in the presence of kinematics parameter errors.

Abstract

We present in-hand manipulation tasks where a robot moves an object in grasp, maintains its external contact mode with the environment, and adjusts its in-hand pose simultaneously. The proposed manipulation task leads to complex contact interactions which can be very susceptible to uncertainties in kinematic and physical parameters. Therefore, we propose a robust in-hand manipulation method, which consists of two parts. First, an in-gripper mechanics model that computes a naïve motion cone assuming all parameters are precise. Then, a robust planning method refines the motion cone to maintain desired contact mode regardless of parametric errors. Real-world experiments were conducted to illustrate the accuracy of the mechanics model and the effectiveness of the robust planning framework in the presence of kinematics parameter errors.
Paper Structure (11 sections, 11 equations, 10 figures, 1 table)

This paper contains 11 sections, 11 equations, 10 figures, 1 table.

Figures (10)

  • Figure 1: We present in-hand manipulation using an extrinsic line contact (shown by the red line in the initial snapshot). The robot only has a rough estimation of the contact line position. Our proposed robust planning successfully pivots the peg while adjusting its in-hand pose and thus, can successfully perform the desired insertion task. A naïve planning approach loses the extrinsic line contact and thus, can not correct the pose of the grasped object. This leads to failure of the plan as seen in the picture.
  • Figure 2: In-hand pivoting with single external line contact. The task is approximated into a 2D contact problem.
  • Figure 3: (a) The grey ellipsoid is the LS in (\ref{['eq:model-LS']}). The two straight pink rays and their spanned wrenches are respectively frictional wrenches $\mathbf{w}_1$, $\mathbf{w}_2$ and $WS$ in (\ref{['eq:model-wrench-cone']}). The solid red arc is the intersection arc in (\ref{['eq:model-wrench-arc']}). The short blue line at the origin (so short that is a little hard to see) is the gravitational wrench. (b) WMS and EMS are respectively the yellow surface and the blue ray. The two green plane and the yellow surface enclosed $WMS\bigoplus EMS$, and its intersection with $\omega=0$ plane is $\mathbf{\Theta}_g$, shown by the deep grey wedge.
  • Figure 4: Case when $\mathbf{\Theta}_g$ is empty. $WMS\bigoplus EMS$ does not intersect with $\omega=0$ plane, meaning it is infeasible to maintain the desired contact mode.
  • Figure 5: Case when $\mathbf{\Theta}_g$ is nonempty and linear cone approximation fails. When (\ref{['eq:model-wrench-arc']}) is a superior arc, the motion cone generated from linear cone approximation (red wedge) flips to the other side. Most of its elements have positive y-motion, which directly lifts up the object, so the red wedge gives an incorrect motion cone.
  • ...and 5 more figures