Dynamically Feasible Deep Reinforcement Learning Policy for Robot Navigation in Dense Mobile Crowds.
Utsav PatelNithish KumarAdarsh Jagan SathyamoorthyDinesh ManochaPublished in: CoRR (2020)
Keyphrases
- robot navigation
- continuous state
- policy search
- reinforcement learning
- optimal policy
- autonomous mobile robot
- markov decision process
- action selection
- autonomous robots
- state space
- mobile devices
- function approximation
- policy gradient
- reward function
- function approximators
- policy iteration
- markov decision processes
- partially observable
- control policies
- action space
- initially unknown
- temporal difference
- landmark recognition
- control policy
- average reward
- map building
- scene understanding
- learning algorithm
- agent learns
- model free
- reinforcement learning algorithms
- long run
- state dependent
- partially observable markov decision processes
- infinite horizon
- finite state
- three dimensional
- computer vision