A novel approach for constructing basis functions in approximate dynamic programming for feedback control.
Jian WangZhenhua HuangXin XuPublished in: ADPRL (2013)
Keyphrases
- feedback control
- approximate dynamic programming
- basis functions
- factored mdps
- optimal control
- dynamic programming
- closed loop
- linear program
- reinforcement learning
- linear combination
- state space
- adaptive control
- step size
- average cost
- radial basis function
- control policy
- multistage
- markov decision processes
- matching pursuit
- real time
- policy iteration
- set of basis functions
- state variables
- np hard