Research map / Reinforcement learning / Visual and representation RL
PlasticineLab: A Soft-Body Manipulation Benchmark with Differentiable Physics
Zhiao Huang, Yuanming Hu, Tao Du, Siyuan Zhou, Hao Su, Joshua B. Tenenbaum, 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 |
|---|---|---|---|
| optimize open-loop action sequences with differentiable physics and Adam | PlasticineLab gradient-based planning · Move, Tri. Move, Rope, RollingPin, Chopsticks, Assembly | normalized incremental IoU | Improved |
| use Adam instead of gradient descent with momentum (ablation) | PlasticineLab gradient-based planning · Move, Tri. Move, Rope, Pinch, RollingPin, Chopsticks, Assembly | normalized incremental IoU: Move: 0.51\pm 0.39 -> 0.90\pm 0.12; Tri. Move: 0.24\pm 0.17 -> 0.35\pm 0.20; Rope: 0.41\pm 0.17 -> 0.59\pm 0.13; Pinch: | Mixed |
| use softened contact rather than hard contact to compute gradients (ablation) | Adam trajectory optimization · most PlasticineLab tasks | normalized incremental IoU | Improved |
| increase yield stress (ablation) | SAC on Move · Move | reward | Improved |
| increase yield stress above 100 (ablation) | SAC on Move · Move | reward | No clear effect |
| increase manipulators and plasticine boxes in the task | RL algorithms · Tri. Move | normalized incremental IoU | Hurt |
| use gradient-based trajectory optimization instead of RL | PlasticineLab planning · Rope, Chopsticks, Assembly | normalized incremental IoU | Improved |
| apply local gradient-based optimization to tasks requiring detachment and reattachment or multiple stages | gradient-based planning · Pinch and Writer | normalized incremental IoU | Hurt |
Closest papers by meaning
- Stabilizing Reinforcement Learning in Differentiable Multiphysics Simulation (ICLR 2025 · similarity 0.89)
- 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 0.87)
- Kinetix: Investigating the Training of General Agents through Open-Ended Physics-Based Control Tasks (ICLR 2025 · similarity 0.87)
- DIFFTACTILE: A Physics-based Differentiable Tactile Simulator for Contact-rich Robotic Manipulation (ICLR 2024 · similarity 0.86)
- DittoGym: Learning to Control Soft Shape-Shifting Robots (ICLR 2024 · similarity 0.86)
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.
