AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

MotionGRPO: Overcoming Low Intra-Group Diversity in GRPO-Based Egocentric Motion Recovery

Nanjie Yao, Junlong Ren, Wenhao Shen, Hao Wang

ICML 2026 · Reinforcement learning › RL from feedback · approach: RL fine-tuning of generative models · 0 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 vanilla GRPO post-training with joint rewards (ablation)EgoAllo diffusion baseline · AMASSMPJPE (mm): 124.985 -> 117.418Improved
add visual reward from perceptual model (ablation)Vanilla GRPO · AMASSGround penetration (m): 0.999 -> 0.886Improved
add visual reward (ablation)Vanilla GRPO · AMASSMPJPE (mm): 117.418 -> 116.549Improved
add Perlin noise injection (ablation)Vanilla GRPO + Visual · AMASSMPJPE (mm): 116.549 -> 114.207Improved
add Perlin noise injection (ablation)Vanilla GRPO + Visual · RICHMPJPE (mm): 189.103 -> 187.223Improved
add Perlin noise injection (ablation)Vanilla GRPO + Visual · AMASSJitter: 2.014 -> 2.000No clear effect
increase Perlin noise intensity from 0 to 0.05 (ablation)GRPO sampling · Training subsetdiversity: 1.8827 -> 2.2048Improved
increase Perlin noise intensity from 0.05 to 0.1 (ablation)GRPO sampling · Training subsetdiversity: 2.2048 -> 3.1430Improved
hard-negative samples instead of GT-noise negatives (ablation)Perceptual model · AMASS perceptual modelaccuracy (%): 86.04 -> 97.68Improved
GRPO instead of direct fine-tuningEgoAllo · AMASSMPJPE (mm): 124.980 -> 114.207Improved
GRPO instead of DPOEgoAllo · AMASSMPJPE (mm): 118.043 -> 114.207Improved
DPO versus baseline on ground penetrationEgoAllo · AMASSGround penetration (m): 1.143 -> 1.530Hurt

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