Continuous-Discrete Reinforcement Learning for Hybrid Control in Robotics.
Michael NeunertAbbas AbdolmalekiMarkus WulfmeierThomas LampeJost Tobias SpringenbergRoland HafnerFrancesco RomanoJonas BuchliNicolas HeessMartin A. RiedmillerPublished in: CoRR (2020)
Keyphrases
- reinforcement learning
- robot control
- control problems
- continuous state spaces
- perception action
- discrete space
- continuous action
- optimal control
- continuous variables
- continuous domains
- continuous state and action spaces
- robot behavior
- discrete variables
- action space
- robotic systems
- control strategies
- control system
- continuous space
- state space
- receding horizon
- machine learning
- industrial robots
- multi agent
- control policy
- inverted pendulum
- artificial intelligence
- computer vision
- unstructured environments
- control method
- control strategy
- action selection
- control policies
- continuous state
- markov decision processes
- real robot
- bayesian networks
- function approximation
- reinforcement learning algorithms
- adaptive control