Adaptively Calibrated Critic Estimates for Deep Reinforcement Learning.
Nicolai DorkaTim WelscheholdJoschka BödeckerWolfram BurgardPublished in: IEEE Robotics Autom. Lett. (2023)
Keyphrases
- reinforcement learning
- function approximation
- actor critic
- temporal difference
- reinforcement learning algorithms
- policy gradient
- model free
- state space
- optimal control
- markov decision processes
- function approximators
- temporal difference learning
- stereo camera
- learning algorithm
- optimal policy
- approximate dynamic programming
- natural actor critic
- policy search
- multi agent
- multi agent systems
- uncalibrated cameras
- deep learning
- markov decision process
- learning classifier systems
- finite state
- estimation error