Skip to content

RopeDreamer: A Kinematic Recurrent State Space Model for Dynamics of Flexible Deformable Linear Objects

Home / Papers / RopeDreamer: A Kinematic Recurrent State Space Model for Dynamics of Flexible Deformable Linear Objects

RopeDreamer: A Kinematic Recurrent State Space Model for Dynamics of Flexible Deformable Linear Objects

Authors: Tim Missal, Lucas Domingues, Berk Guler, Simon Manschitz, Jan Peters, Paula Dornhofer Paro Costa Date: 2026-04-30 Paper ID: openalex:2604.28161

Summary

RopeDreamer is a latent dynamics framework for the robotic manipulation of Deformable Linear Objects (DLOs) that addresses the high-dimensional, non-linear dynamics and physical integrity challenges. By utilizing a Quaternionic Kinematic Chain representation instead of Cartesian positions, the model inherently preserves link-length constancy and topological integrity during complex deformations. A novel dual-decoder architecture further improves accuracy by decoupling state reconstruction from trajectory forecasting, yielding significant improvements in both prediction error and inference efficiency over long-horizon tasks.

Key Contributions

  • Proposes RopeDreamer, a latent dynamics model for DLOs using a Quaternionic Kinematic Chain representation to ensure physical link-length constancy.
  • Introduces a dual-decoder architecture that decouples state reconstruction from future-state prediction to enforce physical awareness in latent dynamics.
  • Achieves a 40.52% reduction in open-loop prediction error over 50-step horizons and 31.17% faster inference compared to current state-of-the-art DLO dynamics models.

Open Questions & Future Work

Key Concepts

Archivist Review

The approved concept provides a robust, reusable geometric constraint for modeling flexible linear structures that avoids physical artifacts like link stretching. The approved open questions address two foundational challenges in robotics and dynamics modeling: the information bottleneck in latent reconstruction and the domain shift in deployment, both of which are high-impact research areas. All other candidates were omitted to keep the knowledge vault focused on high-level mechanisms and foundational research problems.

Approved Concepts

  • Quaternionic Kinematic Chain Representation: It solves the physical constraint problem of link stretching in DLO modeling by moving from Cartesian coordinates to a rotation-based manifold.

Approved Open Questions

  • Hierarchical Latent Architecture Development: The paper explicitly identifies an initial reconstruction trade-off caused by the information bottleneck of the latent manifold, which negatively impacts the starting accuracy of their predictions.
  • Online System Identification Adaptation: The authors explicitly mention this as a key research avenue for deploying their model on physical hardware, addressing the limitation that their current model is primarily evaluated in simulation.

Metadata & Links

url
https://arxiv.org/abs/2604.28161
paper_id
2604.28161
paper_source
openalex
domain
robotics
tags
roboticsdynamics-modelinglatent-space-models
architectures
datasets
concept_slugs
quaternionic-kinematic-chain-representation
dataset_slugs
skill
TimeSeriesSkill
created_at
2026-05-03T06:03:49Z