Research map / Reinforcement learning / Visual and representation RL
D-REX: Differentiable Real-to-Sim-to-Real Engine for Learning Dexterous Grasping
Haozhe Lou, Mingtong Zhang, Haoran Geng, Hanyang Zhou, Sicheng He, Zhiyuan Gao, Siheng Zhao, Jiageng Mao, Pieter Abbeel, Jitendra Malik, Daniel Seita, Yue Wang
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 |
|---|---|---|---|
| condition the grasping policy on object mass that mismatches the evaluation object (ablation) | force-aware grasping policy · object grasping, identical geometry, three masses | success rate: 80 -> 40 and 30 (%) | Hurt |
| condition the policy on identified mass instead of ground-truth mass (ablation) | force-aware grasping policy · object grasping | success rate | No clear effect |
| use force-aware hybrid policy instead of DexGraspNet 2.0 and Human2Sim2Robot | grasping policy · eight-object tabletop grasping | success rate | Improved |
| increase object mass | baseline grasping policies · eight-object tabletop grasping | success rate | Hurt |
| simulate with optimized mass instead of an incorrect lighter mass (ablation) | differentiable engine · pushing trajectory fidelity | trajectory match | Improved |
| identify mass with the differentiable engine across diverse objects | D-REX · mass identification | percentile error: 4.8% to 12.0% error | Improved |
Closest papers by meaning
- Rapidly Adapting Policies to the Real-World via Simulation-Guided Fine-Tuning (ICLR 2025 · similarity 0.87)
- Generalizable Domain Adaptation for Sim-and-Real Policy Co-Training (NeurIPS 2025 · similarity 0.87)
- ASID: Active Exploration for System Identification in Robotic Manipulation (ICLR 2024 · similarity 0.87)
- DexFlyWheel: A Scalable and Self-improving Data Generation Framework for Dexterous Manipulation (NeurIPS 2025 · similarity 0.87)
- Robots Pre-train Robots: Manipulation-Centric Robotic Representation from Large-Scale Robot Datasets (ICLR 2025 · similarity 0.86)
- UniGraspTransformer: Simplified Policy Distillation for Scalable Dexterous Robotic Grasping (CVPR 2025 · 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.
