Cable Manipulation with a Tactile-Reactive Gripper

Yu She, Shaoxiong Wang, Siyuan Dong, Neha Sunil, Alberto Rodriguez, Edward Adelson

Seminal October 2019

Cable following on a braided cable held at one end by the fixed gripper.

TL;DR A gripper with GelSight fingertips follows a free-hanging cable by sliding along it, using only touch to control grip force and cable pose. With an LQR pose controller it follows 98.2% of the length of five different cables, and it inserts a headphone plug in 85% of 20 trials.

What they do

Most cable manipulation keeps the cable quasistatic or constrains it with tables, pegs or clamps. Here a free-hanging cable is followed in real time: a gripper fixed to the table holds its start, and a second gripper on a UR5 slides along it to the tail end, regrasping by feeding the cable back to the fixed gripper when it is about to lose it. Only touch is used, no vision. Following reaches 6.5 cm/s.

How it works

Gripper. A parallel gripper driven by one servo through a tendon and slider-crank, with a flexure parallel-guiding mechanism that replaces 28 rigid parts with one. GelSight fingertips; grip force is controlled at 60 Hz.
Perception. Depth from the GelSight image, a thresholded contact region and PCA give the cable’s offset and angle in the grip. Mean marker displacement approximates the friction force; contact area serves as a grasp-quality check.
Grip controller. A PD controller holds the mean marker displacement at a target; a leaky integrator raises the target when the imprint is too small.
Pose controller. A linear model of the cable state (offset, angle, pulling angle) is fit to about 2000 samples from one cable, collected with a noisy P controller. LQR on this model chooses the pulling direction.

Results

Open-loop motion follows only 36% of the cable. Adding emergency regrasps finishes the task but needs many regrasps; LQR needs the fewest of the four controllers, with little gain in speed.
Fit at 0.025 m/s, the controller still works at 0.045 and 0.065 m/s, with more regrasps per distance but less total time.
Across five cables and ropes it follows 98.2% of total length; the thin, light nylon rope needs the most regrasps.
The robot follows a headphone cable to the plug, detects the plug by touch and inserts it. Insertion alone, with the plug handed over in a random pose, succeeds in 85% of 20 trials.

Limitations

The tactile signal runs at 30 Hz and the gripper loop at 60 Hz. The convex GelSight surface makes it hard to pull the cable back once it reaches the finger edge. The dynamics model is linear, the cable’s start is held by a fixed gripper, and the jack position is calibrated in advance.

Videos

The arm-mounted gripper slides along the cable away from the fixed gripper.

GelSight view of the estimated cable pose and marker displacement as a hand pulls the cable.

Figures

From Figure 1: Following a cable with (a) human hands and (b) robotic grippers.

From Figure 2: Design concept of the cable manipulation system; the tactile imprint gives a depth image and the in-ha... — From Figure 2: Design concept of the cable manipulation system; the tactile imprint gives a depth image and the in-hand cable pose.

From Figure 8: Cable following at three instants: (a) camera view; (b) cable pose estimated from tactile imprints, wi... — From Figure 8: Cable following at three instants: (a) camera view; (b) cable pose estimated from tactile imprints, with the yellow line marking the desired alignment; (c) top view of the end-effector trajectory with the LQR velocity output in red and the pulling angle alpha as a green dotted line.

From Figure 9: Headphone cable following and insertion: (a)(b) following the cable to the plug, (c) plug above the ho... — From Figure 9: Headphone cable following and insertion: (a)(b) following the cable to the plug, (c) plug above the hole with a pose mismatch, (d) plug pose adjusted to align with the hole, (e)(f) plug inserted into the phone's headphone jack.

Abstract

Cables are complex, high dimensional, and dynamic objects. Standard approaches to manipulate them often rely on conservative strategies that involve long series of very slow and incremental deformations, or various mechanical fixtures such as clamps, pins or rings. We are interested in manipulating freely moving cables, in real time, with a pair of robotic grippers, and with no added mechanical constraints. The main contribution of this paper is a perception and control framework that moves in that direction, and uses real-time tactile feedback to accomplish the task of following a dangling cable. The approach relies on a vision-based tactile sensor, GelSight, that estimates the pose of the cable in the grip, and the friction forces during cable sliding. We achieve the behavior by combining two tactile-based controllers: 1) Cable grip controller, where a PD controller combined with a leaky integrator regulates the gripping force to maintain the frictional sliding forces close to a suitable value; and 2) Cable pose controller, where an LQR controller based on a learned linear model of the cable sliding dynamics keeps the cable centered and aligned on the fingertips to prevent the cable from falling from the grip. This behavior is possible by a reactive gripper fitted with GelSight-based high-resolution tactile sensors. The robot can follow one meter of cable in random configurations within 2-3 hand regrasps, adapting to cables of different materials and thicknesses. We demonstrate a robot grasping a headphone cable, sliding the fingers to the jack connector, and inserting it. To the best of our knowledge, this is the first implementation of real-time cable following without the aid of mechanical fixtures.

Summaries are written automatically from each paper's text. Figures and videos come from the paper and its project page; all credit goes to the authors.

Matt Strong