Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Hirschowitz, Ethan, Ramos, Fabio
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2511.09923
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866908650912612352
author Hirschowitz, Ethan
Ramos, Fabio
author_facet Hirschowitz, Ethan
Ramos, Fabio
contents Improving competent robot policies with on-policy RL is often hampered by noisy, low-signal gradients. We revisit Evolution Strategies (ES) as a policy-gradient proxy and localize exploration with bounded, antithetic triangular perturbations, suitable for policy refinement. We propose Triangular-Distribution ES (TD-ES) which pairs bounded triangular noise with a centered-rank finite-difference estimator to deliver stable, parallelizable, gradient-free updates. In a two-stage pipeline - PPO pretraining followed by TD-ES refinement - this preserves early sample efficiency while enabling robust late-stage gains. Across a suite of robotic manipulation tasks, TD-ES raises success rates by 26.5% relative to PPO and greatly reduces variance, offering a simple, compute-light path to reliable refinement.
format Preprint
id arxiv_https___arxiv_org_abs_2511_09923
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Harnessing Bounded-Support Evolution Strategies for Policy Refinement
Hirschowitz, Ethan
Ramos, Fabio
Machine Learning
Artificial Intelligence
Robotics
Improving competent robot policies with on-policy RL is often hampered by noisy, low-signal gradients. We revisit Evolution Strategies (ES) as a policy-gradient proxy and localize exploration with bounded, antithetic triangular perturbations, suitable for policy refinement. We propose Triangular-Distribution ES (TD-ES) which pairs bounded triangular noise with a centered-rank finite-difference estimator to deliver stable, parallelizable, gradient-free updates. In a two-stage pipeline - PPO pretraining followed by TD-ES refinement - this preserves early sample efficiency while enabling robust late-stage gains. Across a suite of robotic manipulation tasks, TD-ES raises success rates by 26.5% relative to PPO and greatly reduces variance, offering a simple, compute-light path to reliable refinement.
title Harnessing Bounded-Support Evolution Strategies for Policy Refinement
topic Machine Learning
Artificial Intelligence
Robotics
url https://arxiv.org/abs/2511.09923