Provably Safe Deep Reinforcement Learning for Robotic Manipulation in Human Environments.
Jakob ThummMatthias AlthoffPublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- human operators
- robot behavior
- robotic systems
- manipulation tasks
- function approximation
- real time
- real world
- mobile robot
- reinforcement learning algorithms
- dynamic environments
- physical world
- human subjects
- real robot
- worst case
- machine learning
- markov decision processes
- optimal policy
- imitation learning
- human behavior
- human robot interaction
- multi agent environments
- reinforcement learning methods
- unstructured environments
- model free
- human interaction
- autonomous robots
- human experts
- human activities
- supervised learning
- dynamic programming
- multi agent
- artificial intelligence
- learning algorithm
- neural network