ExoPredicator: Learning Abstract Models of Dynamic Worlds for Robot Planning
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914394403766272 |
|---|---|
| author | Liang, Yichao Nguyen, Dat Yang, Cambridge Li, Tianyang Tenenbaum, Joshua B. Rasmussen, Carl Edward Weller, Adrian Tavares, Zenna Silver, Tom Ellis, Kevin |
| author_facet | Liang, Yichao Nguyen, Dat Yang, Cambridge Li, Tianyang Tenenbaum, Joshua B. Rasmussen, Carl Edward Weller, Adrian Tavares, Zenna Silver, Tom Ellis, Kevin |
| contents | Long-horizon embodied planning is challenging because the world does not only change through an agent's actions: exogenous processes (e.g., water heating, dominoes cascading) unfold concurrently with the agent's actions. We propose a framework for abstract world models that jointly learns (i) symbolic state representations and (ii) causal processes for both endogenous actions and exogenous mechanisms. Each causal process models the time course of a stochastic cause-effect relation. We learn these world models from limited data via variational Bayesian inference combined with LLM proposals. Across five simulated tabletop robotics environments, the learned models enable fast planning that generalizes to held-out tasks with more objects and more complex goals, outperforming a range of baselines. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_26255 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | ExoPredicator: Learning Abstract Models of Dynamic Worlds for Robot Planning Liang, Yichao Nguyen, Dat Yang, Cambridge Li, Tianyang Tenenbaum, Joshua B. Rasmussen, Carl Edward Weller, Adrian Tavares, Zenna Silver, Tom Ellis, Kevin Artificial Intelligence Computer Vision and Pattern Recognition Machine Learning Robotics Long-horizon embodied planning is challenging because the world does not only change through an agent's actions: exogenous processes (e.g., water heating, dominoes cascading) unfold concurrently with the agent's actions. We propose a framework for abstract world models that jointly learns (i) symbolic state representations and (ii) causal processes for both endogenous actions and exogenous mechanisms. Each causal process models the time course of a stochastic cause-effect relation. We learn these world models from limited data via variational Bayesian inference combined with LLM proposals. Across five simulated tabletop robotics environments, the learned models enable fast planning that generalizes to held-out tasks with more objects and more complex goals, outperforming a range of baselines. |
| title | ExoPredicator: Learning Abstract Models of Dynamic Worlds for Robot Planning |
| topic | Artificial Intelligence Computer Vision and Pattern Recognition Machine Learning Robotics |
| url | https://arxiv.org/abs/2509.26255 |