Active Domain Randomization
Explore this paper's citation graph
Summary
This work empirically examines the effects of domain randomization on agent generalization and proposes Active Domain Randomization, a novel algorithm that learns a parameter sampling strategy that leads to more robust, consistent policies.
- Type
- preprint
- Published
- 2019-04-09
- Cited by
- 352
- References
- 36
- Access
- Open access
- OpenAlex
- https://openalex.org/W2938834661
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:104291994
Keywords
Randomization, Generalization, Computer science, Domain (mathematical analysis), Variance (accounting)
References
- Back to reality: Crossing the reality gap in evolutionary robotics
- Once More Unto the Breach1: Co-evolving a robot and its simulator
- Modeling Purposeful Adaptive Behavior with the Principle of Maximum Causal Entropy
- Bullet physics simulation
- A Tutorial on Bayesian Optimization of Expensive Cost Functions, with Application to Active User Modeling and Hierarchical Reinforcement Learning
- Reinforcement Learning: An Introduction
- MuJoCo: A physics engine for model-based control
- Active learning: theory and applications
- Poppy : open-source, 3D printed and fully-modular robotic platform for science, art and education. (Poppy : plate-forme robotique open source, imprimée en 3D et totalement modulaire pour l'experimentation scientifique, artistique et pédagogique)
- Once More Unto the Breach: Co-evolving a robot and its simulator
- Overcoming catastrophic forgetting in neural networks
- (CAD)^2RL: Real Single-Image Flight without a Single Real Image
- Domain randomization for transferring deep neural networks from simulation to the real world
- Stein Variational Policy Gradient
- Sim-to-Real Transfer of Robotic Control with Dynamics Randomization
- Diversity is All You Need: Learning Skills without a Reward Function
- Which Training Methods for GANs do actually Converge?
- Composable Deep Reinforcement Learning for Robotic Manipulation
- Generalization and Regularization in DQN
- Learning To Simulate
Cited by
- Synthetic Data Generation for Deep Learning-based Semantic Segmentation
- Real-World Robotic Perception and Control Using Synthetic Data
- Synthetic Data for Deep Learning
- ROBEL: Robotics Benchmarks for Learning with Low-Cost Robots
- Solving Rubik's Cube with a Robot Hand
- Robust Domain Randomization for Reinforcement Learning
- Robo-PlaNet: Learning to Poke in a Day
- Neural Data Server: A Large-Scale Search Engine for Transfer Learning Data
- Generating Automatic Curricula via Self-Supervised Active Domain Randomization
- Curriculum in Gradient-Based Meta-Reinforcement Learning
- Rapidly Adaptable Legged Robots via Evolutionary Meta-Learning
- Robust Visual Domain Randomization for Reinforcement Learning.
- Learning to Collide: An Adaptive Safety-Critical Scenarios Generating Method
- Bayesian Domain Randomization for Sim-to-Real Transfer
- Automatic Curriculum Learning For Deep RL: A Short Survey
- Extrapolation in Gridworld Markov-Decision Processes
- UAV detection with a dataset augmented by domain randomization
- Meta-Reinforcement Learning for Robotic Industrial Insertion Tasks
- Multi-Agent Manipulation via Locomotion using Hierarchical Sim2Real
- Learning Active Task-Oriented Exploration Policies for Bridging the Sim-to-Real Gap