Bilevel Planning with Learned Symbolic Abstractions from Interaction Data

Fatih Dogangun, Burcu Kilic, Serdar Bahar, Emre Ugur · 2026

Read Bilevel Planning with Learned Symbolic Abstractions from Interaction Data by Fatih Dogangun, Burcu Kilic, Serdar Bahar, Emre Ugur (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-2603-08599.txt.

Description

Fatih Dogangun and colleagues plan robot moves on two levels, with learned symbols on top and checked motion below. Past symbol learners used fixed rules, skipped plan checks, and missed fine motion detail. Their bilevel frame lets probable symbol rules sketch fast plans while learned motion models test them. It beats symbol-only rivals, matches full forward search, and settles most tasks by fast symbol reasoning.

Catalog record

CREATORFatih Dogangun, Burcu Kilic, Serdar Bahar, Emre Ugur
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE40704
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.08599

Browse the full library · About the Tower of Babel