# Robot learning preprint reports reusable skills with 91.1% success across eight physical tasks

_Published Thursday, November 5, 2026 at 2:05 AM EST · Science, Robotics · Latest · Tier 2 — Notable_

Researchers report in a preprint that PyRUA-REAL, a framework for turning robots' physical interactions into reusable executable skills, achieved an average success rate of 91.1% across eight real-world tasks. The resulting programs can run without an online coding agent making decisions.

The framework lets coding agents retain interaction experience as persistent memory and executable skills, then use execution feedback to refine code and guide subsequent exploration. It also achieved an average success rate of 84.4% on 42 tasks from the RoboDojo benchmark. The physical evaluation centered on manipulation; the broader task evaluation used that benchmark.

## Sources

- [arXiv Query: search_query=cat:cs.RO&id_list=&start=0&max_results=30](https://arxiv.org/abs/2611.06768v1)

---
Canonical: https://techandbusiness.org/newswire/c0zDvpjYU3XHRQ4jTMeeTL
Published: 2026-11-05T07:05:43.495Z
Story chronology: 2026-11-04T16:46:03.000Z
Retrieved: 2026-11-05T10:42:31.254Z
Publisher: Tech & Business (techandbusiness.org)