Magnetic Field-Based Reward Shaping for Goal-Conditioned Reinforcement Learning.
Hongyu DingYuanze TangQing WuBo WangChunlin ChenZhi WangPublished in: CoRR (2023)
Keyphrases
- magnetic field
- reward shaping
- reinforcement learning
- reinforcement learning algorithms
- complex domains
- electromagnetic fields
- agent learns
- state space
- eddy current
- resonant frequency
- function approximation
- markov decision problems
- markov decision processes
- temporal difference
- reward function
- complex environments
- cellular automata
- learning algorithm
- decision theoretic
- action selection
- policy gradient
- policy search
- linear programming
- dynamic programming