Interactive Robotic Moving Cable Segmentation by Motion Correlation
Ondrej Holesovsky, Radoslav Skoviera, Vaclav Hlavac
AI summary
Problem
Passive vision methods struggle to segment individual cables in cluttered, occluded environments due to uniform appearance and complex intersections. Existing interactive approaches either assume only one cable moves or require precise robot arm segmentation masks, limiting practical use.
Approach
The method moves a grasped cable in multiple directions while correlating the gripper's proprioceptive motion with dense optical flow to distinguish the target cable from perturbed neighbors, all without needing arm masks. A grasp sampling algorithm then suggests new interaction points to iteratively expand the segmentation.
Key results
- Motion correlation (MCor) method segments grasped cables despite neighboring cable perturbations
- Novel grasp sampling algorithm proposes new grasp points to improve segmentation recall
- Reformulated problem formulation eliminates dependency on robot arm segmentation masks
- Outperforms the MSeg baseline on a newly recorded dataset of 66 physical robotic sequences
Why it matters
Enables robots to reliably perceive and manipulate tangled deformable objects in complex, real-world environments where passive vision fails.
Abstract
Manipulating tangled hoses, cables, or ropes can be challenging for both robots and humans. Humans often approach these perceptually demanding tasks by pushing or pulling tangled cables and observing the resulting motions. We follow a similar idea to aid robotic cable manipulation. In this letter, we integrate visual and proprioceptive perception to segment a grasped cable by moving it even when the robot or the grasped cable sometimes perturb neighboring cables. We formulate the cable interactive segmentation problem in such a way that our methods do not require robot arm segmentation masks. Furthermore, a novel grasp sampling method can propose new cable grasp points given a partial cable segmentation to improve the segmentation via additional cable-robot interaction. We evaluate the proposed motion correlation (MCor) method on data sequences recorded by our physical robotic setup and show that the method outperforms an earlier motion segmentation (MSeg) baseline.