Image-Based Conditioning for Action Policy Smoothness in Autonomous Miniature Car Racing with Reinforcement Learning.
Bo-Jiun HsuHoang-Giang CaoI LeeChih-Yu KaoJin-Bo HuangI-Chen WuPublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- action selection
- action space
- state action
- agent learns
- optimal policy
- partially observable domains
- car racing
- state space
- autonomous learning
- state and action spaces
- markov decision process
- partially observable
- policy search
- markov decision processes
- temporal difference
- reward shaping
- transition model
- control policies
- continuous state
- control policy
- policy evaluation
- reinforcement learning algorithms
- partially observable environments
- approximate dynamic programming
- reinforcement learning problems
- cost function
- markov decision problems
- multi agent
- partially observable markov decision processes
- reward function
- average reward
- cooperative
- discounted reward
- actor critic
- function approximators
- agent receives
- function approximation
- action models
- reinforcement learning methods
- model free
- prior information
- policy gradient
- inverse reinforcement learning
- finite state
- learning capabilities
- joint action
- initial state
- policy iteration
- dynamic programming
- optimal control
- reward signal
- infinite horizon
- possibility theory
- belief state