Task

class panda_gym.envs.core.Task(sim: PyBullet)

Base class for tasks. :param sim: Simulation instance. :type sim: PyBullet

abstract compute_reward(achieved_goal: ndarray, desired_goal: ndarray, info: Dict[str, Any] = {}) ndarray

Compute reward associated to the achieved and the desired goal.

abstract get_achieved_goal() ndarray

Return the achieved goal.

get_goal() ndarray

Return the current goal.

abstract get_obs() ndarray

Return the observation associated to the task.

abstract is_success(achieved_goal: ndarray, desired_goal: ndarray, info: Dict[str, Any] = {}) ndarray

Returns whether the achieved goal match the desired goal.

abstract reset() None

Reset the task: sample a new goal.