rcognita.controllers.RLController
- class rcognita.controllers.RLController(*args, critic_period=0.1, actor=None, critic=None, time_start=0, action_bounds=None, **kwargs)
Reinforcement learning controller class. Takes instances of actor and critic to operate. Action computation is sampled, i.e., actions are computed at discrete, equi-distant moments in time. critic in turn is updated every critic_period units of time.
- __init__(*args, critic_period=0.1, actor=None, critic=None, time_start=0, action_bounds=None, **kwargs)
Initialize self. See help(type(self)) for accurate signature.
Methods
__init__
(*args[, critic_period, actor, …])Initialize self.
compute_action
(time, observation[, …])compute_action_sampled
(time, observation[, …])reset
(time_start)Resets agent for use in multi-episode simulation.