gym_os2r.tasks¶

gym_os2r.tasks.monopod¶

class gym_os2r.tasks.monopod.MonopodTask(agent_rate, **kwargs)¶

Bases: gym_ignition.base.task.Task, abc.ABC

Monopod task defines the main task functionality for the monopod environment. Task must be wrapped in a runtime or randomizer class to use with igntion or the real robot.

Required *kwargs

Task requires the kwargs; ‘task_mode’, ‘reward_class’, ‘reset_positions’.

Type: dict

task_mode¶

The defined monopod task. current default tasks, ‘free_hip’, ‘fixed_hip’, ‘fixed’, ‘old-free_hip’, ‘old-fixed_hip’, ‘old-fixed’.

Type: str

reward_class¶

Class defining the reward. Must have same functions as RewardBase.

Type: gym_os2r.rewards.RewardBase

reset_positions¶

Reset locations of the task. currently supports, ‘stand’, ‘half_stand’, ‘ground’, ‘lay’, ‘float’.

Type: [str]

observation_index¶

dictionry with the joint_name_pos and joint_name_vel as keys with values corresponding to its index in the observation space.

Type: dict

calculate_reward(obs, action)¶

Calculates the reward given observation and action. The reward is calculated in a provided reward class defined in the tasks kwargs.

Parameters

obs (np.array) – numpy array with the same size task dimensions as observation space.
Deque[np.array] (actions) – Deque of actions taken by the environment numpy array with the same size task dimensions as action space.

Returns

True for done, False otherwise.

Return type

(bool)

create_spaces()¶

Constructs observtion and action spaces for monopod task. Spaces definition is defined in ../config/default/settings.yaml …

Returns: action space. (ndarray): observation space.
Return type: (ndarray)

get_info()¶: Return the info dictionary. :rtype: Dict :returns: A dict with extra information of the task.

get_observation()¶

Returns the current observation state of the monopod.

Returns: Array of joint positions and velocities.
Return type: (ndarray)

get_reward()¶

Returns the reward for the current monopod state.

Returns: True for done, False otherwise.
Return type: (bool)

get_state_info(obs, actions)¶

Returns the reward and is_done for some observation and action space.

Parameters

obs (np.array) – numpy array with the same size task dimensions as observation space.
Deque[np.array] (actions) – Deque of actions taken by the environment numpy array with the same size task dimensions as action space.

Returns

Rewrd given the state. (bool): True for done, False otherwise.

Return type

(Reward)

is_done()¶

Checks if the current state of the robot is outside of the reset_space. logs the reason for the reset as a debug message.

Returns: True for done, False otherwise.
Return type: (bool)

reset_task()¶

Resets the environment into default state. sets the scenario backend into force controller mode Sets the max generalized force for eachcontrolled joint.

Return type: None

set_action(action, store_action=True)¶

Set generalized force target for each controlled joint.

Parameters

action (ndrray) – Generalized force target for each controlled joint.
store_action (bool) – True to store action taken in action history otherwise false to ignore.

Returns

True if success otherwise false.

Return type

(bool)

Raises

(RuntimeError) – Failed to set joints torque target.

gym_os2r.tasks.monopod_no_norm¶

class gym_os2r.tasks.monopod_no_norm.MonopodTask(agent_rate, **kwargs)¶

Bases: gym_ignition.base.task.Task, abc.ABC

Monopod task defines the main task functionality for the monopod environment. Task must be wrapped in a runtime or randomizer class to use with igntion or the real robot.

Required *kwargs

Task requires the kwargs; ‘task_mode’, ‘reward_class’, ‘reset_positions’.

Type: dict

task_mode¶

The defined monopod task. current default tasks, ‘free_hip’, ‘fixed_hip’, ‘fixed’, ‘old-free_hip’, ‘old-fixed_hip’, ‘old-fixed’.

Type: str

reward_class¶

Class defining the reward. Must have same functions as RewardBase.

Type: gym_os2r.rewards.RewardBase

reset_positions¶

Reset locations of the task. currently supports, ‘stand’, ‘half_stand’, ‘ground’, ‘lay’, ‘float’.

Type: [str]

observation_index¶

dictionry with the joint_name_pos and joint_name_vel as keys with values corresponding to its index in the observation space.

Type: dict

calculate_reward(obs, action)¶

Calculates the reward given observation and action. The reward is calculated in a provided reward class defined in the tasks kwargs.

Parameters

obs (np.array) – numpy array with the same size task dimensions as observation space.
Deque[np.array] (actions) – Deque of actions taken by the environment numpy array with the same size task dimensions as action space.

Returns

True for done, False otherwise.

Return type

(bool)

create_spaces()¶

Constructs observtion and action spaces for monopod task. Spaces definition is defined in ../config/default/settings.yaml …

Returns: action space. (ndarray): observation space.
Return type: (ndarray)

get_info()¶: Return the info dictionary. :rtype: Dict :returns: A dict with extra information of the task.

get_observation()¶

Returns the current observation state of the monopod.

Returns: Array of joint positions and velocities.
Return type: (ndarray)

get_reward()¶

Returns the reward for the current monopod state.

Returns: True for done, False otherwise.
Return type: (bool)

get_state_info(obs, actions)¶

Returns the reward and is_done for some observation and action space.

Parameters

obs (np.array) – numpy array with the same size task dimensions as observation space.
Deque[np.array] (actions) – Deque of actions taken by the environment numpy array with the same size task dimensions as action space.

Returns

Rewrd given the state. (bool): True for done, False otherwise.

Return type

(Reward)

is_done()¶

Checks if the current state of the robot is outside of the reset_space. logs the reason for the reset as a debug message.

Returns: True for done, False otherwise.
Return type: (bool)

reset_task()¶

Resets the environment into default state. sets the scenario backend into force controller mode Sets the max generalized force for eachcontrolled joint.

Return type: None

set_action(action, store_action=True)¶

Set generalized force target for each controlled joint.

Parameters

action (ndrray) – Generalized force target for each controlled joint.
store_action (bool) – True to store action taken in action history otherwise false to ignore.

Returns

True if success otherwise false.

Return type

(bool)

Raises

(RuntimeError) – Failed to set joints torque target.