Distributionally Robust Policy Learning via Adversarial Environment Generation

Our goal is to train control policies that generalize well to unseen environments. Inspired by the Distributionally Robust Optimization (DRO) framework, we propose DRAGEN - Distributionally Robust policy learning via Adversarial Generation of ENvironments - for iteratively improving robustness of po...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE robotics and automation letters Ročník 7; číslo 2; s. 1379 - 1386
Hlavní autoři: Ren, Allen Z., Majumdar, Anirudha
Médium: Journal Article
Jazyk:angličtina
Vydáno: Piscataway IEEE 01.04.2022
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Témata:
continual learning
Costs
data sets for robot learning
generalization
Grasping
Optimization
Policies
Reinforcement learning
Robots
Robustness
Task analysis
Training
ISSN:2377-3766, 2377-3766
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Our goal is to train control policies that generalize well to unseen environments. Inspired by the Distributionally Robust Optimization (DRO) framework, we propose DRAGEN - Distributionally Robust policy learning via Adversarial Generation of ENvironments - for iteratively improving robustness of policies to realistic distribution shifts by generating adversarial environments. The key idea is to learn a generative model for environments whose latent variables capture cost-predictive and realistic variations in environments. We perform DRO with respect to a Wasserstein ball around the empirical distribution of environments by generating realistic adversarial environments via gradient ascent on the latent space. We demonstrate strong Out-of-Distribution (OoD) generalization in simulation for (i) swinging up a pendulum with onboard vision and (ii) grasping realistic 3D objects. Grasping experiments on hardware demonstrate better sim2real performance compared to domain randomization.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2377-3766
2377-3766
DOI:10.1109/LRA.2021.3139949