What Matters for Simulation to Online Reinforcement Learning on Real Robots
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866915812904796160 |
|---|---|
| author | As, Yarden Tirumala, Dhruva Zurbrügg, René Li, Chenhao Coros, Stelian Krause, Andreas Wulfmeier, Markus |
| author_facet | As, Yarden Tirumala, Dhruva Zurbrügg, René Li, Chenhao Coros, Stelian Krause, Andreas Wulfmeier, Markus |
| contents | We investigate what specific design choices enable successful online reinforcement learning (RL) on physical robots. Across 100 real-world training runs on three distinct robotic platforms, we systematically ablate algorithmic, systems, and experimental decisions that are typically left implicit in prior work. We find that some widely used defaults can be harmful, while a set of robust, readily adopted design choices within standard RL practice yield stable learning across tasks and hardware. These results provide the first large-sample empirical study of such design choices, enabling practitioners to deploy online RL with lower engineering effort. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_20220 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | What Matters for Simulation to Online Reinforcement Learning on Real Robots As, Yarden Tirumala, Dhruva Zurbrügg, René Li, Chenhao Coros, Stelian Krause, Andreas Wulfmeier, Markus Robotics Artificial Intelligence We investigate what specific design choices enable successful online reinforcement learning (RL) on physical robots. Across 100 real-world training runs on three distinct robotic platforms, we systematically ablate algorithmic, systems, and experimental decisions that are typically left implicit in prior work. We find that some widely used defaults can be harmful, while a set of robust, readily adopted design choices within standard RL practice yield stable learning across tasks and hardware. These results provide the first large-sample empirical study of such design choices, enabling practitioners to deploy online RL with lower engineering effort. |
| title | What Matters for Simulation to Online Reinforcement Learning on Real Robots |
| topic | Robotics Artificial Intelligence |
| url | https://arxiv.org/abs/2602.20220 |