Flow-based Domain Randomization for Learning and Sequencing Robotic Skills
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866916721405722624 |
|---|---|
| author | Curtis, Aidan Li, Eric Noseworthy, Michael Gothoskar, Nishad Chitta, Sachin Li, Hui Kaelbling, Leslie Pack Carey, Nicole |
| author_facet | Curtis, Aidan Li, Eric Noseworthy, Michael Gothoskar, Nishad Chitta, Sachin Li, Hui Kaelbling, Leslie Pack Carey, Nicole |
| contents | Domain randomization in reinforcement learning is an established technique for increasing the robustness of control policies trained in simulation. By randomizing environment properties during training, the learned policy can become robust to uncertainties along the randomized dimensions. While the environment distribution is typically specified by hand, in this paper we investigate automatically discovering a sampling distribution via entropy-regularized reward maximization of a normalizing-flow-based neural sampling distribution. We show that this architecture is more flexible and provides greater robustness than existing approaches that learn simpler, parameterized sampling distributions, as demonstrated in six simulated and one real-world robotics domain. Lastly, we explore how these learned sampling distributions, combined with a privileged value function, can be used for out-of-distribution detection in an uncertainty-aware multi-step manipulation planner. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_01800 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Flow-based Domain Randomization for Learning and Sequencing Robotic Skills Curtis, Aidan Li, Eric Noseworthy, Michael Gothoskar, Nishad Chitta, Sachin Li, Hui Kaelbling, Leslie Pack Carey, Nicole Robotics Artificial Intelligence Machine Learning Domain randomization in reinforcement learning is an established technique for increasing the robustness of control policies trained in simulation. By randomizing environment properties during training, the learned policy can become robust to uncertainties along the randomized dimensions. While the environment distribution is typically specified by hand, in this paper we investigate automatically discovering a sampling distribution via entropy-regularized reward maximization of a normalizing-flow-based neural sampling distribution. We show that this architecture is more flexible and provides greater robustness than existing approaches that learn simpler, parameterized sampling distributions, as demonstrated in six simulated and one real-world robotics domain. Lastly, we explore how these learned sampling distributions, combined with a privileged value function, can be used for out-of-distribution detection in an uncertainty-aware multi-step manipulation planner. |
| title | Flow-based Domain Randomization for Learning and Sequencing Robotic Skills |
| topic | Robotics Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2502.01800 |