Ecological Velocity Planning through Signalized Intersections: A Deep Reinforcement Learning Approach.
Andrea PozziSangjae BaeYongkeun ChoiFrancesco BorrelliDavide Martino RaimondoScott J. MouraPublished in: CDC (2020)
Keyphrases
- reinforcement learning
- action selection
- planning problems
- partially observable
- deterministic domains
- state space
- function approximation
- machine learning
- model free
- multi agent
- learning algorithm
- temporal difference
- stochastic domains
- decision support
- partial observability
- single agent
- complex domains
- reinforcement learning algorithms
- dynamic programming
- markov decision processes
- ai planning
- policy search
- domain independent
- supervised learning
- heuristic search
- partially observable markov decision processes
- markov decision problems
- reinforcement learning problems
- neural network
- velocity vector
- blocks world
- mixed initiative
- plan execution
- goal oriented
- velocity field
- learning process
- optical flow
- transfer learning