FISAR: Forward Invariant Safe Reinforcement Learning with a Deep Neural Network-Based Optimizer.
Chuangchuang SunDong-Ki KimJonathan P. HowPublished in: ICRA (2021)
Keyphrases
- reinforcement learning
- neural network
- function approximation
- state space
- reinforcement learning algorithms
- supervised learning
- robotic control
- affine invariant
- affine transformation
- query optimization
- optimal control
- multi agent reinforcement learning
- optimization algorithm
- transition model
- multi agent
- invariance properties
- bi directional
- learning problems
- model free
- temporal difference
- database
- reinforcement learning methods
- temporal difference learning
- markov decision process
- computer vision
- action selection
- invariant features
- e learning
- learning process
- markov decision processes