def step(self):
if self._prevact is None:
self.start()
rot = self._env.step(self._prevact)
roat = Reward_observation_action_terminal()
roat.o = rot.o
roat.r = rot.r
roat.terminal = rot.terminal
self._total_reward += rot.r
if rot.terminal == 1:
self._episode_count += 1
converged = self._agent.end(rot.r)
roat.converged = converged if converged is not None else 0
self._prevact = None
else:
self._step_count += 1
self._prevact = self._agent.step(rot.r, rot.o)
roat.a = self._prevact
return roat
