AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Rethinking Bimanual Robotic Manipulation: Learning with Decoupled Interaction Framework

Jian-Jian Jiang, Xiao-Ming Wu, Yi-Xiang He, Ling-An Zeng, Yi-Lin Wei, Dandan Zhang, Wei-Shi Zheng

ICCV 2025 · Reinforcement learning › Visual and representation RL · approach: Dexterous manipulation transfer · 16 citations (OpenAlex)

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 changedWhereMeasuredEffect
add decoupled design (independent model per arm) (ablation)integrated control baseline · RoboTwin ablation (Flow Matching, 7 tasks)average success rate: 0.426 -> 0.624Improved
add selective scaling factors on top of decoupled design (ablation)decoupled framework · RoboTwin ablation (Flow Matching, 7 tasks)average success rate: 0.624 -> 0.754Improved
add selective bias vectors on top of decoupled design (ablation)decoupled framework · RoboTwin ablation (Flow Matching, 7 tasks)average success rate: 0.624 -> 0.747Improved
add both scaling factors and bias vectors (ablation)decoupled framework · RoboTwin ablation (Flow Matching, 7 tasks)average success rate: 0.624 -> 0.789Improved
replace MLP interaction with selective interaction (ablation)decoupled framework · RoboTwin (7 tasks)average success rate: 0.736 -> 0.789Improved
replace concatenation interaction with selective interaction (ablation)decoupled framework · RoboTwin (7 tasks)average success rate: 0.733 -> 0.789Improved
apply decoupled interaction to PFMPoint Flow Matching · RoboTwin integration into existing methodsaverage success rate: 0.426 -> 0.789Improved
apply decoupled interaction to DP3 (DDIM)DP3 · RoboTwin integration into existing methodsaverage success rate: 0.554 -> 0.674Improved
enlarge DP3 action generator from 262.43M to 576.43MDP3 · RoboTwin model sizeaverage success rate: 0.554 -> 0.361Hurt
shrink our model from 536.07M to 42.95M parametersDecoupled Interaction Framework · RoboTwin model sizeaverage success rate: 0.789 -> 0.719No clear effect
increase demonstrations from 150 to 200Decoupled Interaction Framework · RoboTwin Blocks Stackaverage success rate: 0.530 -> 0.650Improved
use decoupled interaction instead of DP3 in real worldDP3 · Real-world (4 tasks, 60 trials)success count: 27/60 -> 45/60Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy