AdaReP:Adaptive Re-Planning under Model Mismatch for Neural World-Model Predictive Control
2026-06-22 • Robotics
RoboticsArtificial Intelligence
AI summaryⓘ
The authors study how robots plan their actions step-by-step using predictions about the world, which can be slow because they replan so often. They look into how reusing old plans can save time but might cause errors depending on how much the situation has changed. To balance this, they introduce AdaReP, a tool that decides when to replan based on how different the current situation is from the predicted one, without changing the original robot model. Their tests show AdaReP cuts down planning time a lot while still performing well, even in real robot tasks.
neural world modelsmodel predictive controlreplanningdynamic regretlocal dynamics sensitivitycached plan reuseperturbation analysisrobotic manipulationlatent-space controlplanner computation overhead
Authors
Yutian Cheng, Xiaojian Ma, Xianhao Wang, Min Yang, Rongpeng Su, Hangxin Liu, Xi Chen, Shuai Li, Qing Li
Abstract
Neural world models coupled with model predictive control (MPC) replan at every environment step to bound accumulated prediction error, but this incurs substantial computational overhead. Reusing a cached plan reduces this overhead, yet its effectiveness depends on how prediction mismatch propagates through the local dynamics. We analyze this trade-off with a perturbation-based dynamic-regret framework and show that stale-plan penalties scale with the reuse tolerance, the accumulated mismatch since the last replanning step, and the local dynamics sensitivity. Based on this structure, we propose AdaReP, a training-free wrapper that adapts the replanning tolerance online using the current deviation from the cached rollout and a local sensitivity estimate, without modifying the learned world model or planner. Across image-space planning, latent-space control, and real-world robotic manipulation, AdaReP substantially reduces planner-side computation while maintaining comparable task performance, including over 80% fewer queries on a 50-trial physical robot study.