Disentangling Dense Multi-Cable Knots

Viswanath, Vainavi; Grannen, Jennifer; Sundaresan, Priya; Thananjeyan, Brijen; Novoseller, Ellen; Ichnowski, Jeffrey; Laskey, Michael; Goldberg, Ken

doi:10.1109/iros51168.2021.9636397

Cited by 17 publications

(21 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Predicting task-relevant keypoints using fullyconvolutional neural networks is a popular technique in deformable manipulation applications such as fabric smoothing [12,38], t-shirt folding [12,23], and cable untangling [15,42,45]. Seita et al [38] and Ganapathi et al [12] both predict keypoints corresponding to the corners of a rectangular towel to implement a smoothing policy that iteratively pulls identified corners away from the towel.…”

Section: A Towel Corner Detection For Smoothing and Foldingmentioning

confidence: 99%

See 1 more Smart Citation

All You Need is LUV: Unsupervised Collection of Labeled Images using Invisible UV Fluorescent Indicators

Thananjeyan¹,

Kerr²,

Huang³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Large-scale semantic image annotation is a significant challenge for learning-based perception systems in robotics. Current approaches often rely on human labelers, which can be expensive, or simulation data, which can visually or physically differ from real data. This paper proposes Labels from UltraViolet (LUV), a novel framework that enables rapid, labeled data collection in real manipulation environments without human labeling. LUV uses transparent, ultravioletfluorescent paint with programmable ultraviolet LEDs to collect paired images of a scene in standard lighting and UV lighting to autonomously extract segmentation masks and keypoints via color segmentation. We apply LUV to a suite of diverse robot perception tasks to evaluate its labeling quality, flexibility, and data collection rate. Results suggest that LUV is 180-2500 times faster than a human labeler across the tasks. We show that LUV provides labels consistent with human annotations on unpainted test images. The networks trained on these labels are used to smooth and fold crumpled towels with 83% success rate and achieve 1.7mm position error with respect to human labels on a surgical needle pose estimation task. The low cost of LUV makes it ideal as a lightweight replacement for human labeling systems, with the one-time setup costs at $300 equivalent to the cost of collecting around 200 semantic segmentation labels on Amazon Mechanical Turk. Code, datasets, visualizations, and supplementary material can be found at https://sites. google.com/berkeley.edu/luv.

show abstract

Section: A Towel Corner Detection For Smoothing and Foldingmentioning

confidence: 99%

“…Prior work in cable manipulation [15,42,45] often assumes that the cable is visually distinguishable from the background via color segmentation. But cables in household or industrial settings may not be chromatically distinct from background objects.…”

Section: B Cable Segmentationmentioning

confidence: 99%

All You Need is LUV: Unsupervised Collection of Labeled Images using Invisible UV Fluorescent Indicators

Thananjeyan¹,

Kerr²,

Huang³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Caged pull-apart actions can loosen and undo knots using learning-based perception to identify isolated knots in the scene. The caged pull apart action slides apart from a knot center, avoiding the need to pinch and regrasp the cable repeatedly to undo knots as in prior work [20,3,22]. Building on our previous work, this paper contributes a learning-based perception pipeline which leverages RGBD sensing to identify cable endpoints and knots [22,20,3], which are used to select and parameterize the novel manipulation primitives.…”

Section: Introductionmentioning

confidence: 99%

“…Cables-defined here as single-dimensional deformable objects-often tangle and form knots, which can be unsightly, unsafe, and reduce utility. However, autonomously manipulating cables to untangle them is challenging due to their infinite-dimensional configuration spaces and tendency to form self-occlusions and knots [3,20,22,19,11]. These challenges grow with increasing cable length as longer cables allow more knots to form.…”

Section: Introductionmentioning

confidence: 99%

“…Prior work in robot cable untangling considers short cables and studies how to untangle dense knots, which have no open space between adjacent cable segments [3,20,22], or considers longer cables, estimated to be no longer than one meter in length, whose paths can be clearly traced using analytic methods [11]. This paper studies untangling knots in cables up to 3 m long that have sufficient stiffness to prevent the formation of dense knots.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Autonomously Untangling Long Cables

Viswanath¹,

Shivakumar²,

Kerr³

et al. 2022

Robotics: Science and Systems XVIII

Self Cite

View full text Add to dashboard Cite

Cables are ubiquitous in many settings, but are prone to self-occlusions and knots, making them difficult to perceive and manipulate. The challenge often increases with cable length: long cables require more complex slack management and strategies to facilitate observability and reachability. In this paper, we focus on autonomously untangling cables up to 3 meters in length using a bilateral robot. We develop new motion primitives to efficiently untangle long cables and novel gripper jaws specialized for this task. We present Sliding and Grasping for Tangle Manipulation (SGTM), an algorithm that composes these primitives with RGBD vision to iteratively untangle. SGTM untangles cables with success rates of 67% on isolated overhand and figure eight knots and 50% on more complex configurations. Supplementary material, visualizations, and videos can be found at https://sites.google.com/ view/rss-2022-untangling/home.

show abstract