Relay Hindsight Experience Replay: Self-guided continual reinforcement learning for sequential object manipulation tasks with sparse rewards.
Yongle LuoYuxin WangKun DongQiang ZhangErkang ChengZhiyong SunBo SongPublished in: Neurocomputing (2023)
Keyphrases
- reinforcement learning
- object manipulation
- manipulation tasks
- robot control
- robot navigation
- real robot
- markov decision processes
- state space
- function approximation
- action selection
- machine learning
- motion planning
- model free
- high dimensional
- transfer learning
- position and orientation
- human robot interaction
- d objects
- mobile robot
- computer vision
- learning algorithm
- robotic systems
- optimal policy
- optimal control
- dynamic environments
- nonlinear systems
- real world