mlsanityData to 14 Oct 2026

Aim Short to Reach Far: Your Frozen World Model Can Plan Better Than You Think

arXiv 2609.30036 ↗First posted 24 Sep 2026cs.LG cs.ROcited by 0

On the map: World models → World models for navigation also fits: Latent predictive world models

Abstract (arXiv)

Latent world models plan toward goal images with a frozen pretrained predictor, without task rewards or extra trained heads. However, their planners struggle with long-range goals, and prior work addresses this by training extra components such as value functions or subgoal models. We show that the planning target itself can cause this failure: even with exact dynamics and globally optimal short-horizon search, scoring predictions by their distance to the final goal rejects the first steps of a route that initially moves away from the goal. Building on this insight, we propose Anchored Planning (AP), a training-free method that reuses the world model's own offline trajectories. AP retrieves a segment that leads from the current observation toward the goal and aims the frozen planner at an observation shortly after the segment's start. Across four diverse tasks, AP substantially improves frozen LeWM planners for both action synthesis and action ranking, and it outperforms both additional final-goal search and the LeWM planner on long-range goals.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in World models for navigation, one of 11 directions in World models. Builds on 28 of its 37 references that are in the map. First posted on arXiv 24 Sep 2026.

Builds on

28 of 37 references.

Closest ancestors · 1

Same direction, newest first

  1. 2026Latent World Models with Monotone Planning Costs for Image-Goal Navigation10 Aug

Other references · 27

  1. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 2,838
  2. 2018Learning Latent Dynamics for Planning from Pixelscited by 1,264
  3. 2017Hindsight Experience Replaycited by 1,136
  4. 2022Planning with Diffusion for Flexible Behavior Synthesiscited by 1,061
  5. 2023Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecturecited by 1,039
  6. 2025V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planningcited by 809
  7. 2023TD-MPC2: Scalable, Robust World Models for Continuous Controlcited by 512
  8. 2021Implicit Behavioral Cloningcited by 452
  9. 2022Temporal Difference Learning for Model Predictive Controlcited by 397
  10. 2024DINO-WM: World Models on Pre-trained Visual Features enable Zero-shot Planningcited by 386
  11. 2026LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixelscited by 273
  12. 2025LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristicscited by 254
15 more
  1. 2024OGBench: Benchmarking Offline Goal-Conditioned RLcited by 238
  2. 2018Visual Foresight: Model-Based Deep Reinforcement Learning for Vision-Based Robotic Controlcited by 230
  3. 2018Semi-parametric Topological Memory for Navigationcited by 209
  4. 2019Search on the Replay Buffer: Bridging Planning and Reinforcement Learningcited by 194
  5. 2021The Surprising Effectiveness of Representation Learning for Visual Imitationcited by 154
  6. 2021Learning Generalizable Robotic Reward Functions from "In-The-Wild" Human Videoscited by 119
  7. 2025Learning from Reward-Free Offline Data: A Case for Planning with Latent Dynamics Modelscited by 89
  8. 2025What Drives Success in Physical Planning with Joint-Embedding Predictive World Models?cited by 43
  9. 2026Hierarchical Planning with Latent World Modelscited by 43
  10. 2026SAGE: Subgoal-Conditioned Action Generation for Latent World Model Planningcited by 7
  11. 2022PALMER: Perception-Action Loop with Memory for Long-Horizon Planningcited by 5
  12. 2026IMWM: Intuition Models Complement World Models for Latent Planningcited by 4
  13. 2026The Objective Is the Bottleneck: Latent World Models Encode What Their Planners Cannot Usecited by 3
  14. 2026Traj-LeWM: Path-Aware World-Model Planning via Latent Trajectory Costcited by 3
  15. 2025Zero-shot World Models via Search in Memorycited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026How Much Planning Is Enough? Reducing Search and Computation in World-Model Planning
  2. 2026The Planning Limits of Latent World Models
  3. 2026How Long, Not How Close: A Learned Temporal Metric for Planning in Latent World Models
  4. 2026PRISM: PRior-guided Imagination Sampling in world Models
  5. 2026Beyond a single latent space: a dual-latent world model for long-horizon planning
  6. 2026FIRM-WM: State-factorized factual-interventional recurrent modeling for reward-free visual planning
  7. 2026Beyond Visual Quality: A Study of Test-Time Planning with World Action Models
  8. 2026Memory as Plans: World-Action Modeling with Memory-Grounded Planning