AI Paper Atlas

Research map / Vision-language and multimodal / Robot action and navigation

Sparse Imagination for Efficient Visual World Model Planning

Junha Chun, Youngjoon Jeong, Taesup Kim

ICLR 2026 · Vision-language and multimodal › Robot action and navigation · approach: Embodied multimodal LLM agents · 7 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
drop 50% of patch tokens during planning (ablation)DINO-WM planner · PushTplanning time (s/iter): 173 -> 82Improved
drop 50% of patch tokens during planning (ablation)DINO-WM planner · PushTsuccess rate (%): 75.0 -> 70.0Mixed
drop 10-50% tokens (ablation)DINO-WM planner · average of 6 environmentssuccess rate (%): 69.8 -> 69.7No clear effect
drop 90% of tokens (ablation)DINO-WM planner · average of 6 environmentssuccess rate (%): 69.8 -> 46.7Hurt
use sparse patch tokens instead of CLS-only tokenCLS-Token world model · Granularsuccess rate (%): 20.0 -> 85.0Improved
add sparse imagination planning to SmolVLA at 50% dropSmolVLA · LIBERO-10success rate (%): 29 -> 33Improved
use 50% drop instead of full patches (ablation)VLA-guided planner · LIBERO-10time per episode (s): 53.4 -> 29.7Improved
add sparse imagination planning to SmolVLASmolVLA · Meta-World (all)success rate (%): 41.87 -> 47.73Improved
add sparse imagination planning on real robotSmolVLA · Real-world PickPlacesuccess rate (%): 60 -> 80Improved
train with randomized grouped attention (ablation)sparse-token world model · 50% token drop predictionnormalized L2 error: 0.036 -> 0.016Improved
use attention-encoder importance selection instead of random (ablation)sparse imagination · Wall vision-onlysuccess rate (%): 58.3 -> 21.7Hurt
use sophisticated token selection (LTRP, STAR, attention) instead of random (ablation)sparse imagination · average over 6 envssuccess rate (%): 66.7 -> 59.5-64.0No clear effect

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