For questions about RL agents that can perform more than one action at each time step.
A Reinforcement Learning (RL) agent chooses an action among a set of discrete or continuous actions for each time step in the environment. This action selection, given the current state, influences the environment transition.
The prevalence on RL is on policies that map states to a single individual action, or combine multiple low-level action decisions as a single high-level action.
In real-world problems, the agent may need to control multiple actions simultaneously at a given time step. These actions would rarely be independent and the agent's learning would benefit from learning the underlying relationship between the actions.
