Offline Goal-Conditioned Reinforcement Learning with Projective Quasimetric Planning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kobanda, Anthony, Radji, Waris, Petitbois, Mathieu, Maillard, Odalric-Ambrym, Portelas, Rémy
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912862339858432
author Kobanda, Anthony
Radji, Waris
Petitbois, Mathieu
Maillard, Odalric-Ambrym
Portelas, Rémy
author_facet Kobanda, Anthony
Radji, Waris
Petitbois, Mathieu
Maillard, Odalric-Ambrym
Portelas, Rémy
contents Offline Goal-Conditioned Reinforcement Learning seeks to train agents to reach specified goals from previously collected trajectories. Scaling that promises to long-horizon tasks remains challenging, notably due to compounding value-estimation errors. Principled geometric offers a potential solution to address these issues. Following this insight, we introduce Projective Quasimetric Planning (ProQ), a compositional framework that learns an asymmetric distance and then repurposes it, firstly as a repulsive energy forcing a sparse set of keypoints to uniformly spread over the learned latent space, and secondly as a structured directional cost guiding towards proximal sub-goals. In particular, ProQ couples this geometry with a Lagrangian out-of-distribution detector to ensure the learned keypoints stay within reachable areas. By unifying metric learning, keypoint coverage, and goal-conditioned control, our approach produces meaningful sub-goals and robustly drives long-horizon goal-reaching on diverse a navigation benchmarks.
format Preprint
id arxiv_https___arxiv_org_abs_2506_18847
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Offline Goal-Conditioned Reinforcement Learning with Projective Quasimetric Planning
Kobanda, Anthony
Radji, Waris
Petitbois, Mathieu
Maillard, Odalric-Ambrym
Portelas, Rémy
Machine Learning
Offline Goal-Conditioned Reinforcement Learning seeks to train agents to reach specified goals from previously collected trajectories. Scaling that promises to long-horizon tasks remains challenging, notably due to compounding value-estimation errors. Principled geometric offers a potential solution to address these issues. Following this insight, we introduce Projective Quasimetric Planning (ProQ), a compositional framework that learns an asymmetric distance and then repurposes it, firstly as a repulsive energy forcing a sparse set of keypoints to uniformly spread over the learned latent space, and secondly as a structured directional cost guiding towards proximal sub-goals. In particular, ProQ couples this geometry with a Lagrangian out-of-distribution detector to ensure the learned keypoints stay within reachable areas. By unifying metric learning, keypoint coverage, and goal-conditioned control, our approach produces meaningful sub-goals and robustly drives long-horizon goal-reaching on diverse a navigation benchmarks.
title Offline Goal-Conditioned Reinforcement Learning with Projective Quasimetric Planning
topic Machine Learning
url https://arxiv.org/abs/2506.18847