ACID: Action Consistency via Inverse Dynamics for Planning with World Models
About
Decision-time planning with action-conditioned world models has become a popular paradigm for embodied control. However, the standard planning cost judges a candidate solely by how close its predicted terminal state lies to the goal, leaving the realizability of the intermediate transitions unchecked -- a predicted trajectory can look convincing while the environment rollout drifts away from it. In this paper, we propose ACID, a decision-time planning framework that introduces cycle action consistency: the action inferred backward from a predicted transition by an inverse dynamics model should recover the one that was conditioned on. We fold this per-step residual into the planning cost via a scale-invariant adaptive weight. Across four action-conditioned world models and six tasks spanning rigid and deformable manipulation, articulated control, and visual navigation, ACID consistently improves planning and matches the baseline's accuracy with substantially less planning compute.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Table-top manipulation | Push T | Success Rate100 | 9 | |
| Deformable body manipulation | Rope | Chamfer Distance0.56 | 6 | |
| articulated control | reacher | Success Rate90 | 4 | |
| Robotic Arm Manipulation | Cube | Success Rate74 | 4 | |
| Deformable Object Manipulation | Granular | Chamfer Distance0.3 | 2 | |
| Visual Navigation | Goal-conditioned visual navigation | ATE1.2835 | 2 |