Deep Reinforcement Learning with Adaptive Hierarchical Reward for MultiMulti-Phase Multi Multi-Objective Dexterous Manipulation.
Lingfeng TaoJiucai ZhangXiaoli ZhangPublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- multi objective
- multi objective optimization
- optimization algorithm
- manipulation tasks
- evolutionary algorithm
- function approximation
- learning algorithm
- genetic algorithm
- eligibility traces
- learning process
- hierarchical reinforcement learning
- multi objective optimization problems
- objective function
- reinforcement learning algorithms
- partially observable environments
- dynamic programming
- optimal policy
- multi agent
- haptic feedback
- reward function
- particle swarm optimization
- multiple layers
- reward shaping
- learning capabilities
- machine learning
- multiple objectives
- nsga ii
- humanoid robot
- state space
- multi objective evolutionary
- total reward
- agent receives
- average reward
- multi objective evolutionary algorithms
- partially observable
- temporal difference
- action selection