I am reading this paper on battery management using RL. The action consist in the charging/discharging power of the battery at timestep $t$. For instance, in the case of the charging power, the maximum of this action can be given by the maximum charging speed $c^{\max }$ or by the state of charge of the battery, since it cannot be charged more than $100\%$. Therefore, the charging action has the following range:

$$ 0 \leq c_{t} \leq \min \left\{c^{\max }, \frac{B^{\max }-B_{t}}{\eta_{c}}\right\} $$

In some timesteps the maximum of $c_{t}$ will be $c^{\max }$ and for others $\frac{B^{\max }-B_{t}}{\eta_{c}}$. What would be the best way of implementing a variable action range? I have thought in using a range $[0,1]$ for the action, scaling it to the suitable range. Is there any standard way to deal with variable ranges?.

  • $\begingroup$ Here is a related question, but your question seems to be in the context of continuous action spaces, so I would recommend that you change your title to emphasize that. See also this. $\endgroup$
    – nbro
    Commented Feb 28, 2022 at 10:16


You must log in to answer this question.