Bilevel Learning for Bilevel Planning
Bowen Li 1 2 , Tom Silver 3 , Sebastian Scherer 1 , and Alexander Gray 2 · 2025
Read Bilevel Learning for Bilevel Planning by Bowen Li 1 2 , Tom Silver 3 , Sebastian Scherer 1 , and Alexander Gray 2 (2025) 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-2502-08697.txt.
Description
Bowen Li and co-authors teach robots to learn the very concepts they plan with, instead of receiving them from engineers. Bilevel planning lets a robot reason in abstract relational terms, then ground each step in motion. Its power depends on predicates, the learned or hand-made concepts like 'holding' or 'clear'. Hand-made ones do not scale to messy, high-dimensional worlds.
Their system, IVNTR, learns neural predicates straight from demonstrations. Symbolic learning of what each predicate does alternates with neural learning of how to recognize it, each guiding the other. A robot that invents its own planning vocabulary from watching could generalize chores to new kitchens, tools, and tasks.
Catalog record
| CREATOR | Bowen Li 1 2 , Tom Silver 3 , Sebastian Scherer 1 , and Alexander Gray 2 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 78 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-08697