Self-Correcting Visual Planning Method Uses Frozen World Models with Uncertainty Awareness
A new paper proposes a visual planner combining a frozen latent world model with sampling-based model predictive control, designed to self-correct when executed actions depart from predictions at deployment.
Read the full story at arXiv cs.RO ↗
Timeline · 1 report
- 2026-11-19 05:00 · arXiv cs.RO
How Uncertain Is My Action? Self-Correcting Visual Planning with Frozen World Models