Implicit Kinematic Policies: Unifying Joint and Cartesian Action Spaces in End-to-End Robot Learning.
Aditya GanapathiPete FlorenceJake VarleyKaylee BurnsKen GoldbergAndy ZengPublished in: ICRA (2022)
Keyphrases
- end to end
- action space
- control policies
- state space
- optimal policy
- markov decision processes
- markov decision process
- continuous state
- markov decision problems
- real valued
- reinforcement learning
- state and action spaces
- admission control
- stochastic processes
- policy search
- skill learning
- congestion control
- continuous action
- degrees of freedom
- heuristic search
- partially observable markov decision processes
- action selection
- application layer
- scalable video
- mobile robot
- single agent
- partially observable
- dynamical systems
- random variables
- probability distribution
- dynamic programming