Learning Domain Randomization Distributions for Training Robust Locomotion Policies

2019-06-02 · Melissa Mozifian, Juan Camilo Gamboa Higuera, David Meger, Gregory Dudek

Domain randomization (DR) is a successful technique for learning robust policies for robot systems, when the dynamics of the target robot system are unknown. The success of policies trained with domain randomization however, is highly dependent on the correct selection of the randomization distribution. The majority of success stories typically use real world data in order to carefully select the DR distribution, or incorporate real world trajectories to better estimate appropriate randomization distributions. In this paper, we consider the problem of finding good domain randomization parameters for simulation, without prior access to data from the target system. We explore the use of gradient-based search methods to learn a domain randomization with the following properties: 1) The trained policy should be successful in environments sampled from the domain randomization distribution 2) The domain randomization distribution should be wide enough so that the experience similar to the target robot system is observed during training, while addressing the practicality of training finite capacity models. These two properties aim to ensure the trajectories encountered in the target system are close to those observed during training, as existing methods in machine learning are better suited for interpolation than extrapolation. We show how adapting the domain randomization distribution while training context-conditioned policies results in improvements on jump-start and asymptotic performance when transferring a learned policy to the target environment.

📄 PDF Abstract BibTeX arXiv:1906.00410

Code (0)

등록된 구현이 없습니다.

Learning and Deploying Robust Locomotion Policies with Minimal Dynamics Randomization

2022-09-26 · Luigi Campanaro, Siddhant Gangapurwala, Wolfgang Merkt, Ioannis Havoutis

Training deep reinforcement learning (DRL) locomotion policies often require massive amounts of data to converge to the desired behaviour. In this regard, simulators provide a cheap and abundant source. For successful si…

Deep Reinforcement Learning Domain Adaptation

Sim-to-Real of Humanoid Locomotion Policies via Joint Torque Space Perturbation Injection

2026-03-23 · Junhyeok Rui Cha, Woohyun Cha, Jaeyong Shin, Donghyeon Kim 외 arxiv

This paper proposes a novel alternative to existing sim-to-real methods for training control policies with simulated experiences. Unlike prior methods that typically rely on domain randomization over a fixed finite set o…

Learning Sim-to-Real Humanoid Locomotion in 15 Minutes

2025-12-01 · Younggyo Seo, Carmelo Sferrazza, Juyue Chen, Guanya Shi 외 arxiv

Massively parallel simulation has reduced reinforcement learning (RL) training time for robots from days to minutes. However, achieving fast and reliable sim-to-real RL for humanoid control remains difficult due to the c…

Reinforcement Learning

The Role of Domain Randomization in Training Diffusion Policies for Whole-Body Humanoid Control

2024-11-02 · Oleg Kaidanov, Firas Al-Hafez, Yusuf Suvari, Boris Belousov 외

Humanoids have the potential to be the ideal embodiment in environments designed for humans. Thanks to the structural similarity to the human body, they benefit from rich sources of demonstration data, e.g., collected vi…

Diversity Humanoid Control

Reinforcement Learning for Robust Parameterized Locomotion Control of Bipedal Robots

2021-03-26 · Zhongyu Li, Xuxin Cheng, Xue Bin Peng, Pieter Abbeel 외

Developing robust walking controllers for bipedal robots is a challenging endeavor. Traditional model-based locomotion controllers require simplifying assumptions and careful modelling; any small errors can result in uns…

reinforcement-learning Reinforcement Learning Reinforcement Learning (RL)