Slot-MPC: Goal-Conditioned Model Predictive Control with Object-Centric Representations
Jonathan Spieler, Angel Villar-Corrales, Sven Behnke · 2026
Read Slot-MPC: Goal-Conditioned Model Predictive Control with Object-Centric Representations by Jonathan Spieler, Angel Villar-Corrales, Sven Behnke (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-14937.txt.
Description
Jonathan Spieler, Angel Villar-Corrales, and Sven Behnke present Slot-MPC, a planner that reasons about scenes object by object. Vision encoders learn slot representations, one per object, and an action-conditioned model predicts how those objects move. At run time, model predictive control plans actions through the learned model, and gradients make that planning faster than sampling-based rivals. Simulated robot manipulation trials show gains in both task success and planning efficiency.
Catalog record
| CREATOR | Jonathan Spieler, Angel Villar-Corrales, Sven Behnke |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51464 |
| LICENSE | http://creativecommons.org/licenses/by-sa/4.0/ |
| RIGHTS | CC BY SA |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.14937