SWIRL: A sequential windowed inverse reinforcement learning algorithm for robot tasks with delayed rewards.
Sanjay KrishnanAnimesh GargRichard LiawBrijen ThananjeyanLauren MillerFlorian T. PokornyKen GoldbergPublished in: Int. J. Robotics Res. (2019)
Keyphrases
- robotic systems
- service robots
- autonomous robots
- robot control
- mobile robot
- visually guided
- navigation tasks
- robot navigation
- reinforcement learning
- fourier transform
- humanoid robot
- human users
- path planning
- robot behavior
- vision system
- multiple tasks
- manipulation tasks
- unstructured environments
- parallel robot
- goal directed
- real robot
- real time
- machine learning
- neural network