Reproduce “AssistDLO: Assistive Teleoperation for Deformable Linear Object Manipulation” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Berk Guler, Simon Manschitz, Kay Pompetzki, Jan Peters

AssistDLO presents a shared-autonomy system for teleoperating two robot arms during deformable linear object manipulation. By combining wrist-mounted RGB-D rope reconstruction, stereoscopic visual overlays, and a Control Barrier Function controller that guides grasping while avoiding unintended rope contacts, the system reduced successful-task completion time from 115.01 seconds in pure teleoperation to roughly 91–96 seconds with assistance; physical action assistance was especially beneficial for inexperienced users.

Key insights

  • Visual and physical assistance address different bottlenecks: visual overlays improve spatial understanding, whereas SA-CBF helps users execute precise grasps while preserving the surrounding rope configuration.
  • Benefits depend on operator experience and object properties. Novice users gained the clearest success-rate improvement from SA-CBF, while experts generally preferred visual assistance because physical intervention could conflict with their planned motions.
  • Fixed assistance strategies are not uniformly effective: local action guidance helped with shorter or stiffer ropes but was less useful for the long, highly deformable rope, where limited wrist-camera coverage made global visual context more important.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.