AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

Dejavu: Towards Experience Feedback Learning for Embodied Intelligence

Shaokai Wu, Yanbiao Ji, Qiuchang Li, Zhiyi Zhang, Qichen He, Wenyuan Xie, Guodong Zhang, Bayram Bayramli, Yue Ding, Hongtao Lu

CVPR 2026 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Embodied agent learning · 4 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 EFN with a larger experience bank (Vol=1000 vs Vol=300) (ablation)OpenVLA + EFN · LIBERO averagesuccess rate (%): 84.4 -> 87.0Improved
add EFN residual retrieval module to frozen VLAOpenVLA · LIBERO averagesuccess rate (%): 76.5 -> 87.0Improved
add EFN to frozen VLAOpenVLA · LIBERO-Longsuccess rate (%): 53.7 -> 76.5Improved
add EFN to stronger base policyUniVLA · LIBERO averagesuccess rate (%): 95.2 -> 97.2Improved
replace EFN with kNN-RAG (copy action from nearest past step)OpenVLA · LIBERO averagesuccess rate (%): 76.5 -> 73.2Hurt
add retrieval-only kNN-RAG on real robotGO-1 · AgiBot-G1 real worldsuccess rate (%): 25.8 -> 18.2Hurt
add test-time finetuning GC-TTT on backboneGO-1 · AgiBot-G1 real worldsuccess rate (%): 25.8 -> 26.2No clear effect
add residual-only correction without retrieval (ResAct)OpenVLA · LIBERO averagesuccess rate (%): 76.5 -> 79.2Improved
use EFN with frozen backbone instead of R2A backbone updatesGO-1 · AgiBot-G1 real worldsuccess rate (%): 39.5 -> 50.8Improved
add SAC (entropy-regularized RL) for EFN training (ablation)OpenVLA + EFN · LIBEROsuccess rate and stepsImproved
add similarity-shaped dense rewards (ablation)OpenVLA + EFN · LIBEROsuccess rateImproved
add instruction-level retrieval filtering (ablation)OpenVLA + EFN · LIBEROsuccess rateImproved

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