def compute_reward(self, achieved_goal, desired_goal,
info: Dict[str, Any]) -> Union[np.ndarray, float]:
d = distance(achieved_goal, desired_goal)
if self.reward_type == "sparse":
return -np.array(d > self.distance_threshold, dtype=np.float64)
else:
return -d
def is_success(self, achieved_goal: np.ndarray, desired_goal: np.ndarray) -> Union[np.ndarray, float]:
d = distance(achieved_goal, desired_goal)
return np.array(d < self.distance_threshold, dtype=np.float64)