Q-Learning with Recurrent Neural Networks as a Controller for the Inverted Pendulum Problem.
Ahmet OnatHajime KitaYoshikazu NishikawaPublished in: ICONIP (1998)
Keyphrases
- inverted pendulum
- recurrent neural networks
- feedback control
- simulation study
- intelligent control
- nonlinear systems
- adaptive fuzzy
- reinforcement learning
- neural network
- fuzzy controller
- function approximation
- feed forward
- control algorithm
- fuzzy systems
- learning algorithm
- state space
- biped robot
- mobile robot
- echo state networks
- learning rate
- artificial neural networks
- neuro fuzzy
- initial conditions
- legged robots
- recurrent networks
- neural model
- optimal policy
- reinforcement learning algorithms
- adaptive control
- fuzzy control
- knowledge base
- closed loop
- fuzzy logic
- reinforcement learning methods
- real time
- optimal control
- back propagation