Saved in:
Bibliographic Details
Main Authors: Pannacci, Matteo, Fanti, Andrea, Umili, Elena, Capobianco, Roberto
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2602.09761
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910017800634368
author Pannacci, Matteo
Fanti, Andrea
Umili, Elena
Capobianco, Roberto
author_facet Pannacci, Matteo
Fanti, Andrea
Umili, Elena
Capobianco, Roberto
contents In this work we address the problem of training a Reinforcement Learning agent to follow multiple temporally-extended instructions expressed in Linear Temporal Logic in sub-symbolic environments. Previous multi-task work has mostly relied on knowledge of the mapping between raw observations and symbols appearing in the formulae. We drop this unrealistic assumption by jointly training a multi-task policy and a symbol grounder with the same experience. The symbol grounder is trained only from raw observations and sparse rewards via Neural Reward Machines in a semi-supervised fashion. Experiments on vision-based environments show that our method achieves performance comparable to using the true symbol grounding and significantly outperforms state-of-the-art methods for sub-symbolic environments.
format Preprint
id arxiv_https___arxiv_org_abs_2602_09761
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Grounding LTL Tasks in Sub-Symbolic RL Environments for Zero-Shot Generalization
Pannacci, Matteo
Fanti, Andrea
Umili, Elena
Capobianco, Roberto
Machine Learning
Artificial Intelligence
In this work we address the problem of training a Reinforcement Learning agent to follow multiple temporally-extended instructions expressed in Linear Temporal Logic in sub-symbolic environments. Previous multi-task work has mostly relied on knowledge of the mapping between raw observations and symbols appearing in the formulae. We drop this unrealistic assumption by jointly training a multi-task policy and a symbol grounder with the same experience. The symbol grounder is trained only from raw observations and sparse rewards via Neural Reward Machines in a semi-supervised fashion. Experiments on vision-based environments show that our method achieves performance comparable to using the true symbol grounding and significantly outperforms state-of-the-art methods for sub-symbolic environments.
title Grounding LTL Tasks in Sub-Symbolic RL Environments for Zero-Shot Generalization
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2602.09761