Pre-Training for Robots: Offline RL Enables Learning New Tasks in a Handful of Trials.
Aviral KumarAnikait SinghFrederik D. EbertMitsuhiko NakamotoYanlai YangChelsea FinnSergey LevinePublished in: Robotics: Science and Systems (2023)
Keyphrases
- reinforcement learning
- online learning
- learning algorithm
- learning process
- learning speed
- active learning
- multiple tasks
- supervised learning
- autonomous robots
- knowledge acquisition
- robot control
- robotic systems
- learning systems
- learning problems
- humanoid robot
- training process
- learning tasks
- multi task learning
- feedforward neural networks
- mobile robot
- multi agent
- previously learned
- motor skills
- computer based training