Research map / Reinforcement learning / Visual and representation RL
DLO-Lab: Benchmarking Deformable Linear Object Manipulations with Differentiable Physics
Junyi Cao, Yian Wang, Ziyan Xiong, Chunru Lin, Zhehuan Chen, Chuang Gan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use Coefficient mode instead of Candidate mode for VLM grasp proposal (ablation) | CMA-ES · DLO-Lab Unknotting | max episodic return: 57.21 -> 3.06 | Hurt |
| use Marker mode instead of Candidate mode for VLM grasp proposal (ablation) | CMA-ES · DLO-Lab Wrapping | max episodic return: 162.68 -> 136.39 | Hurt |
| use Coefficient mode instead of Candidate mode for VLM grasp proposal (ablation) | CMA-ES · DLO-Lab Wrapping | max episodic return: 162.68 -> 144.78 | Hurt |
| use Coefficient mode instead of Candidate mode for VLM grasp proposal (ablation) | CMA-ES · DLO-Lab Lifting | max episodic return: 335.59 -> 330.57 | Hurt |
| add dynamic plan update to VLM task decomposition (ablation) | DLO agent with CMA-ES · Wiring-ring | task completion (qualitative) | Improved |
| add dynamic plan update to VLM task decomposition (ablation) | DLO agent with CMA-ES · Letter Art | task completion (qualitative) | No clear effect |
| use sample-based CMA-ES trajectory optimization instead of gradient descent | trajectory optimization · DLO-Lab Unknotting | max episodic return: 3.44 -> 57.21 | Improved |
| use first-order model-based RL (SAPO) instead of PPO/SAC | RL · DLO-Lab Unknotting | max episodic return: 3 -> 46 | Improved |
| use CMA-ES instead of SAPO | FO-MBRL · DLO-Lab success rate average | success rate (%): 35.0% -> 86.6% | Improved |
| use GD instead of CMA-ES on smooth tasks | trajectory optimization · DLO-Lab Separation | max episodic return: 84.86 -> 115.52 | Mixed |
Closest papers by meaning
- Thin-Shell Object Manipulations With Differentiable Physics Simulations (ICLR 2024 · similarity 0.87)
- DeformGen: Dynamics-Based Topology Augmentation for Deformable Manipulation Policy Learning (NeurIPS 2026 · similarity 0.85)
- Geometry-aware RL for Manipulation of Varying Shapes and Deformable Objects (ICLR 2025 · similarity 0.85)
- DIFFTACTILE: A Physics-based Differentiable Tactile Simulator for Contact-rich Robotic Manipulation (ICLR 2024 · similarity 0.84)
- RobotSmith: Generative Robotic Tool Design for Acquisition of Complex Manipulation Skills (NeurIPS 2025 · similarity 0.84)
- GarmentLab: A Unified Simulation and Benchmark for Garment Manipulation (NeurIPS 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
