High-level Reward Deep Reinforcement Learning Approach for a Novel Physical-Logical Hybrid Factory Line Robot Vehicle Simulation.
Ryota HigaShinji NakadaiPublished in: CASE (2022)
Keyphrases
- reinforcement learning
- real robot
- high level
- mobile robot
- low level
- simulated robot
- robot behavior
- function approximation
- real environment
- robot control
- experimental platform
- autonomous learning
- state space
- production line
- learning agent
- model free
- temporal difference
- reinforcement learning algorithms
- vision system
- real time
- human robot interaction
- eligibility traces
- physical world
- machine learning
- multi robot
- pedestrian detection
- perceptual aliasing
- long run
- humanoid robot
- markov decision processes
- control policy
- learning algorithm
- agent learns
- robocup soccer
- policy gradient
- physical objects
- robot manipulators
- action space
- reward function
- path planning
- higher level