MultiDLO: Simultaneous Shape Tracking of Multiple Deformable Linear Objects with Global-Local Topology Preservation

TL;DR

MultiDLO is a real-time algorithm that simultaneously tracks multiple deformable linear objects in RGB-D sequences, preserving their topology and identity even when entangled, with demonstrated effectiveness on complex scenarios.

Contribution

It introduces a novel global-local topology preservation method for real-time multi-DLO shape tracking, handling multiple entangled objects simultaneously.

Findings

Successfully tracks multiple entangled DLOs in real-time

Maintains accurate local geometry and object identity

Open-source implementation available for community use

Abstract

MultiDLO is a real-time algorithm for estimating the shapes of multiple, intertwining deformable linear objects (DLOs) from RGB-D image sequences. Unlike prior methods that track only a single DLO, MultiDLO simultaneously handles several objects. It uses the geodesic distance in the Global-Local Topology Preservation algorithm to define both inter-object identity and intra-object topology, ensuring entangled DLOs remain distinct with accurate local geometry. The MultiDLO algorithm is demonstrated on two challenging scenarios involving three entangling ropes, and the implementation is open-source and available for the community.