The Art of Reinforcement Learning: Fundamentals, Mathematics, and Implementations with Python

© The Author(s), under exclusive license to APress Media, LLC, part of Springer Nature 2023

M. HuThe Art of Reinforcement Learninghttps://doi.org/10.1007/978-1-4842-9606-6_10

10. Problems with Continuous Action Space

Michael Hu¹  

(1)

Shanghai, Shanghai, China

 

In the previous chapters, we covered reinforcement learning problems with discrete action spaces, where actions are typically represented using discrete numbers, such as integers. However, in the real world, many problems require continuous actions that cannot be represented by integers. Continuous action spaces are common in real-world applications such as robotics, where precise and continuous control is essential.

This chapter focuses on using policy-based methods to solve reinforcement ...