rofunc.learning.RofuncRL.tasks.isaacgymenv.hotu.tmp

1.  Module Contents

1.1.  Classes

Humanoid	This class is a wrapper of the Isaac Gym environment for the Humanoid task.
HumanoidHOTU

1.2.  Functions

build_amp_observations
dof_to_obs
compute_humanoid_observations
compute_humanoid_observations_max
compute_humanoid_reward
compute_humanoid_reset

1.3.  API

class rofunc.learning.RofuncRL.tasks.isaacgymenv.hotu.tmp.Humanoid(config, rl_device, sim_device, graphics_device_id, headless, virtual_screen_capture, force_render)

Bases: rofunc.learning.RofuncRL.tasks.isaacgymenv.base.vec_task.VecTask

This class is a wrapper of the Isaac Gym environment for the Humanoid task.

Initialization

Initialise the VecTask.

Args:

config: config dictionary for the environment. sim_device: the device to simulate physics on. eg. ‘cuda:0’ or ‘cpu’ graphics_device_id: the device ID to render with. headless: Set to False to disable viewer rendering. virtual_screen_capture: Set to True to allow the users get captured screen in RGB array via env.render(mode=’rgb_array’). force_render: Set to True to always force rendering in the steps (if the control_freq_inv is greater than 1 we suggest stting this arg to True)

get_obs_size()

get_action_size()

get_num_actors_per_env()

create_sim(**kwargs)

reset_idx(env_ids)

set_char_color(col)

pre_physics_step(actions)

post_physics_step()

render(sync_frame_time=False)

class rofunc.learning.RofuncRL.tasks.isaacgymenv.hotu.tmp.HumanoidHOTU(cfg, rl_device, sim_device, graphics_device_id, headless, virtual_screen_capture, force_render)

Bases: rofunc.learning.RofuncRL.tasks.isaacgymenv.hotu.tmp.Humanoid

class StateInit

Bases: enum.Enum

Default = 0

Start = 1

Random = 2

Hybrid = 3

post_physics_step()

get_num_amp_obs()

property amp_observation_space

fetch_amp_obs_demo(num_samples)

build_amp_obs_demo(motion_ids, motion_times0)

reset_idx(env_ids)

rofunc.learning.RofuncRL.tasks.isaacgymenv.hotu.tmp.build_amp_observations(root_pos: Tensor, root_rot: Tensor, root_vel: Tensor, root_ang_vel: Tensor, dof_pos: Tensor, dof_vel: Tensor, key_body_pos: Tensor, local_root_obs: bool, root_height_obs: bool, dof_obs_size: int, dof_offsets: List[int]) → Tensor

rofunc.learning.RofuncRL.tasks.isaacgymenv.hotu.tmp.dof_to_obs(pose: Tensor, dof_obs_size: int, dof_offsets: List[int]) → Tensor

rofunc.learning.RofuncRL.tasks.isaacgymenv.hotu.tmp.compute_humanoid_observations(root_pos: Tensor, root_rot: Tensor, root_vel: Tensor, root_ang_vel: Tensor, dof_pos: Tensor, dof_vel: Tensor, key_body_pos: Tensor, local_root_obs: bool, root_height_obs: bool, dof_obs_size: int, dof_offsets: List[int]) → Tensor

rofunc.learning.RofuncRL.tasks.isaacgymenv.hotu.tmp.compute_humanoid_observations_max(body_pos: Tensor, body_rot: Tensor, body_vel: Tensor, body_ang_vel: Tensor, local_root_obs: bool, root_height_obs: bool) → Tensor

rofunc.learning.RofuncRL.tasks.isaacgymenv.hotu.tmp.compute_humanoid_reward(obs_buf: Tensor) → Tensor

rofunc.learning.RofuncRL.tasks.isaacgymenv.hotu.tmp.compute_humanoid_reset(reset_buf: Tensor, progress_buf: Tensor, contact_buf: Tensor, contact_body_ids: Tensor, rigid_body_pos: Tensor, max_episode_length: float, enable_early_termination: bool, termination_heights: Tensor) → Tuple[Tensor, Tensor]