Continuous-Discrete Reinforcement Learning for Hybrid Control in Robotics.
Michael NeunertAbbas AbdolmalekiMarkus WulfmeierThomas LampeJost Tobias SpringenbergRoland HafnerFrancesco RomanoJonas BuchliNicolas HeessMartin A. RiedmillerPublished in: CoRL (2019)
Keyphrases
- reinforcement learning
- robot control
- perception action
- control problems
- continuous state spaces
- continuous domains
- optimal control
- discrete space
- action space
- continuous variables
- continuous state and action spaces
- continuous action
- continuous state
- continuous space
- artificial intelligence
- computer vision
- robotic systems
- adaptive control
- continuous data
- discrete variables
- function approximation
- discrete geometry
- dynamic programming
- action selection
- industrial robots
- piecewise linear
- unstructured environments
- receding horizon
- discrete data
- control policy
- markov decision processes
- control strategies
- control strategy
- control method
- motor control
- control policies
- control system
- feedback control
- mobile robot
- state space
- robotic control
- machine learning