Language

unilab.envs.locomotion.go2_arm.manip_loco¶

Classes

`ArmStageConfig`	ArmStageConfig(freeze_arm_joints: 'bool' = False, disable_ee_goal_trajectory: 'bool' = False, fixed_ee_goal_cart: 'list[float]' = <factory>)
`CommandsConfig`	CommandsConfig(vel_limit: 'list[list[float]]' = <factory>, resampling_time: 'float' = 0.0, heading_command: 'bool' = False, heading_range: 'list[float]' = <factory>, heading_control_stiffness: 'float' = 0.5, rel_standing_envs: 'float' = 0.0, resample_time_s: 'float \| None' = None, zero_command_prob: 'float' = 0.2)
`CurriculumConfig`	Expand velocity command ranges when mean tracking_lin_vel exceeds a threshold.
`EEGoalConfig`	End-effector goal config in spherical coordinates.
`Go2ArmDomainRandConfig`	Go2ArmDomainRandConfig(randomize_base_mass: 'bool' = False, added_mass_range: 'list[float]' = <factory>, randomize_body_mass: 'bool' = False, body_mass_multiplier_range: 'list[float]' = <factory>, random_com: 'bool' = False, com_offset_x: 'list[float]' = <factory>, randomize_gravity: 'bool' = False, gravity_range: 'list[list[float]]' = <factory>, randomize_ground_friction: 'bool' = False, ground_friction_multiplier_range: 'list[float]' = <factory>, randomize_dof_armature: 'bool' = False, dof_armature_multiplier_range: 'list[float]' = <factory>, push_robots: 'bool' = False, push_interval: 'int' = 750, max_force: 'list[float]' = <factory>, push_body_name: 'str \| None' = None, randomize_kp: 'bool' = True, kp_multiplier_range: 'list[float]' = <factory>, randomize_kd: 'bool' = True, kd_multiplier_range: 'list[float]' = <factory>)
`Go2ArmManipLocoCfg`	Go2ArmManipLocoCfg(scene: 'SceneCfg' = <factory>, sim_dt: 'float' = 0.01, max_episode_seconds: 'float' = 20.0, ctrl_dt: 'float' = 0.02, render_spacing: float = 1.0, render_offset_mode: str = 'grid', motrix_max_iterations: Optional[int] = None, post_step_forward_sensor: bool = False, control_config: 'ControlConfig' = <factory>, noise_config: 'NoiseConfig' = <factory>, sensor: 'Go2ArmSensor' = <factory>, ik: 'IKConfig' = <factory>, asset: 'Asset' = <factory>, iterations: 'int \| None' = None, model_file: 'str' = <factory>, init_state: 'InitState' = <factory>, commands: 'CommandsConfig' = <factory>, reward_config: 'RewardConfig \| None' = None, domain_rand: 'Go2ArmDomainRandConfig' = <factory>, goal_ee: 'EEGoalConfig' = <factory>, history: 'HistoryConfig' = <factory>, arm_stage: 'ArmStageConfig' = <factory>, curriculum: 'CurriculumConfig' = <factory>)
`Go2ArmManipLocoDRProvider`
`Go2ArmManipLocoEnv`
`HistoryConfig`	Actor/critic observation history lengths.
`InitState`	InitState(pos: 'list[float]' = <factory>)
`RewardConfig`	RewardConfig(scales: 'dict[str, float]', tracking_sigma: 'float', base_height_target: 'float', target_foot_height: 'float' = 0.1, object_sigma: 'float' = 0.1, leg_dof_upper_limits: 'list[float]' = <factory>, leg_dof_lower_limits: 'list[float]' = <factory>, dof_pos_limit_margin: 'float' = 0.01)

class unilab.envs.locomotion.go2_arm.manip_loco.InitState[source]¶

Bases: object

InitState(pos: ‘list[float]’ = <factory>)

Parameters:: pos (list[float])

pos: list[float]¶

__init__(pos=<factory>)¶

Parameters:: pos (list[float])

class unilab.envs.locomotion.go2_arm.manip_loco.Go2ArmDomainRandConfig[source]¶

Bases: DomainRandConfig

Go2ArmDomainRandConfig(randomize_base_mass: ‘bool’ = False, added_mass_range: ‘list[float]’ = <factory>, randomize_body_mass: ‘bool’ = False, body_mass_multiplier_range: ‘list[float]’ = <factory>, random_com: ‘bool’ = False, com_offset_x: ‘list[float]’ = <factory>, randomize_gravity: ‘bool’ = False, gravity_range: ‘list[list[float]]’ = <factory>, randomize_ground_friction: ‘bool’ = False, ground_friction_multiplier_range: ‘list[float]’ = <factory>, randomize_dof_armature: ‘bool’ = False, dof_armature_multiplier_range: ‘list[float]’ = <factory>, push_robots: ‘bool’ = False, push_interval: ‘int’ = 750, max_force: ‘list[float]’ = <factory>, push_body_name: ‘str | None’ = None, randomize_kp: ‘bool’ = True, kp_multiplier_range: ‘list[float]’ = <factory>, randomize_kd: ‘bool’ = True, kd_multiplier_range: ‘list[float]’ = <factory>)

Parameters:

randomize_base_mass (bool)
added_mass_range (list[float])
randomize_body_mass (bool)
body_mass_multiplier_range (list[float])
random_com (bool)
com_offset_x (list[float])
randomize_gravity (bool)
gravity_range (list[list[float]])
randomize_ground_friction (bool)
ground_friction_multiplier_range (list[float])
randomize_dof_armature (bool)
dof_armature_multiplier_range (list[float])
push_robots (bool)
push_interval (int)
max_force (list[float])
push_body_name (str | None)
randomize_kp (bool)
kp_multiplier_range (list[float])
randomize_kd (bool)
kd_multiplier_range (list[float])

randomize_kp: bool = True¶

kp_multiplier_range: list[float]¶

randomize_kd: bool = True¶

kd_multiplier_range: list[float]¶

__init__(randomize_base_mass=False, added_mass_range=<factory>, randomize_body_mass=False, body_mass_multiplier_range=<factory>, random_com=False, com_offset_x=<factory>, randomize_gravity=False, gravity_range=<factory>, randomize_ground_friction=False, ground_friction_multiplier_range=<factory>, randomize_dof_armature=False, dof_armature_multiplier_range=<factory>, push_robots=False, push_interval=750, max_force=<factory>, push_body_name=None, randomize_kp=True, kp_multiplier_range=<factory>, randomize_kd=True, kd_multiplier_range=<factory>)¶

Parameters:

randomize_base_mass (bool)
added_mass_range (list[float])
randomize_body_mass (bool)
body_mass_multiplier_range (list[float])
random_com (bool)
com_offset_x (list[float])
randomize_gravity (bool)
gravity_range (list[list[float]])
randomize_ground_friction (bool)
ground_friction_multiplier_range (list[float])
randomize_dof_armature (bool)
dof_armature_multiplier_range (list[float])
push_robots (bool)
push_interval (int)
max_force (list[float])
push_body_name (str | None)
randomize_kp (bool)
kp_multiplier_range (list[float])
randomize_kd (bool)
kd_multiplier_range (list[float])

class unilab.envs.locomotion.go2_arm.manip_loco.EEGoalConfig[source]¶

Bases: object

End-effector goal config in spherical coordinates.

Parameters:

sphere_l_range (list[float])
sphere_phi_range (list[float])
sphere_theta_range (list[float])
traj_time_range (list[float])
hold_time_range (list[float])
collision_upper_limits (list[float])
collision_lower_limits (list[float])
underground_limit (float)
num_collision_check_samples (int)
num_resample_attempts (int)
default_orn_roll (float)
arm_induced_pitch (float)
delta_orn_r (list[float])
delta_orn_p (list[float])
delta_orn_y (list[float])
init_ee_cart (list[float])

sphere_l_range: list[float]¶

sphere_phi_range: list[float]¶

sphere_theta_range: list[float]¶

traj_time_range: list[float]¶

hold_time_range: list[float]¶

collision_upper_limits: list[float]¶

collision_lower_limits: list[float]¶

underground_limit: float = -0.57¶

num_collision_check_samples: int = 10¶

num_resample_attempts: int = 10¶

default_orn_roll: float = 1.5707963267948966¶

arm_induced_pitch: float = 0.78¶

delta_orn_r: list[float]¶

delta_orn_p: list[float]¶

delta_orn_y: list[float]¶

init_ee_cart: list[float]¶

__init__(sphere_l_range=<factory>, sphere_phi_range=<factory>, sphere_theta_range=<factory>, traj_time_range=<factory>, hold_time_range=<factory>, collision_upper_limits=<factory>, collision_lower_limits=<factory>, underground_limit=-0.57, num_collision_check_samples=10, num_resample_attempts=10, default_orn_roll=1.5707963267948966, arm_induced_pitch=0.78, delta_orn_r=<factory>, delta_orn_p=<factory>, delta_orn_y=<factory>, init_ee_cart=<factory>)¶

Parameters:

sphere_l_range (list[float])
sphere_phi_range (list[float])
sphere_theta_range (list[float])
traj_time_range (list[float])
hold_time_range (list[float])
collision_upper_limits (list[float])
collision_lower_limits (list[float])
underground_limit (float)
num_collision_check_samples (int)
num_resample_attempts (int)
default_orn_roll (float)
arm_induced_pitch (float)
delta_orn_r (list[float])
delta_orn_p (list[float])
delta_orn_y (list[float])
init_ee_cart (list[float])

class unilab.envs.locomotion.go2_arm.manip_loco.CommandsConfig[source]¶

Bases: Commands

CommandsConfig(vel_limit: ‘list[list[float]]’ = <factory>, resampling_time: ‘float’ = 0.0, heading_command: ‘bool’ = False, heading_range: ‘list[float]’ = <factory>, heading_control_stiffness: ‘float’ = 0.5, rel_standing_envs: ‘float’ = 0.0, resample_time_s: ‘float | None’ = None, zero_command_prob: ‘float’ = 0.2)

Parameters:

vel_limit (list[list[float]])
resampling_time (float)
heading_command (bool)
heading_range (list[float])
heading_control_stiffness (float)
rel_standing_envs (float)
resample_time_s (float | None)
zero_command_prob (float)

resample_time_s: float | None = None¶

zero_command_prob: float = 0.2¶

__init__(vel_limit=<factory>, resampling_time=0.0, heading_command=False, heading_range=<factory>, heading_control_stiffness=0.5, rel_standing_envs=0.0, resample_time_s=None, zero_command_prob=0.2)¶

Parameters:

vel_limit (list[list[float]])
resampling_time (float)
heading_command (bool)
heading_range (list[float])
heading_control_stiffness (float)
rel_standing_envs (float)
resample_time_s (float | None)
zero_command_prob (float)

class unilab.envs.locomotion.go2_arm.manip_loco.CurriculumConfig[source]¶

Bases: object

Expand velocity command ranges when mean tracking_lin_vel exceeds a threshold.

Parameters:

enable (bool)
threshold (float)
step_size (list[float])
max_vel_limit (list[float])

enable: bool = False¶

threshold: float = 0.8¶

step_size: list[float]¶

max_vel_limit: list[float]¶

__init__(enable=False, threshold=0.8, step_size=<factory>, max_vel_limit=<factory>)¶

Parameters:

enable (bool)
threshold (float)
step_size (list[float])
max_vel_limit (list[float])

class unilab.envs.locomotion.go2_arm.manip_loco.RewardConfig[source]¶

Bases: object

RewardConfig(scales: ‘dict[str, float]’, tracking_sigma: ‘float’, base_height_target: ‘float’, target_foot_height: ‘float’ = 0.1, object_sigma: ‘float’ = 0.1, leg_dof_upper_limits: ‘list[float]’ = <factory>, leg_dof_lower_limits: ‘list[float]’ = <factory>, dof_pos_limit_margin: ‘float’ = 0.01)

Parameters:

scales (dict[str, float])
tracking_sigma (float)
base_height_target (float)
target_foot_height (float)
object_sigma (float)
leg_dof_upper_limits (list[float])
leg_dof_lower_limits (list[float])
dof_pos_limit_margin (float)

scales: dict[str, float]¶

tracking_sigma: float¶

base_height_target: float¶

target_foot_height: float = 0.1¶

object_sigma: float = 0.1¶

leg_dof_upper_limits: list[float]¶

leg_dof_lower_limits: list[float]¶

dof_pos_limit_margin: float = 0.01¶

__init__(scales, tracking_sigma, base_height_target, target_foot_height=0.1, object_sigma=0.1, leg_dof_upper_limits=<factory>, leg_dof_lower_limits=<factory>, dof_pos_limit_margin=0.01)¶

Parameters:

scales (dict[str, float])
tracking_sigma (float)
base_height_target (float)
target_foot_height (float)
object_sigma (float)
leg_dof_upper_limits (list[float])
leg_dof_lower_limits (list[float])
dof_pos_limit_margin (float)

class unilab.envs.locomotion.go2_arm.manip_loco.HistoryConfig[source]¶

Bases: object

Actor/critic observation history lengths. A value of 1 disables history.

Parameters:

num_actor_history (int)
num_critic_history (int)

num_actor_history: int = 1¶

num_critic_history: int = 1¶

__init__(num_actor_history=1, num_critic_history=1)¶

Parameters:

num_actor_history (int)
num_critic_history (int)

class unilab.envs.locomotion.go2_arm.manip_loco.ArmStageConfig[source]¶

Bases: object

ArmStageConfig(freeze_arm_joints: ‘bool’ = False, disable_ee_goal_trajectory: ‘bool’ = False, fixed_ee_goal_cart: ‘list[float]’ = <factory>)

Parameters:

freeze_arm_joints (bool)
disable_ee_goal_trajectory (bool)
fixed_ee_goal_cart (list[float])

freeze_arm_joints: bool = False¶

disable_ee_goal_trajectory: bool = False¶

fixed_ee_goal_cart: list[float]¶

__init__(freeze_arm_joints=False, disable_ee_goal_trajectory=False, fixed_ee_goal_cart=<factory>)¶

Parameters:

freeze_arm_joints (bool)
disable_ee_goal_trajectory (bool)
fixed_ee_goal_cart (list[float])

class unilab.envs.locomotion.go2_arm.manip_loco.Go2ArmManipLocoCfg[source]¶

Bases: Go2ArmBaseCfg

Go2ArmManipLocoCfg(scene: ‘SceneCfg’ = <factory>, sim_dt: ‘float’ = 0.01, max_episode_seconds: ‘float’ = 20.0, ctrl_dt: ‘float’ = 0.02, render_spacing: float = 1.0, render_offset_mode: str = ‘grid’, motrix_max_iterations: Optional[int] = None, post_step_forward_sensor: bool = False, control_config: ‘ControlConfig’ = <factory>, noise_config: ‘NoiseConfig’ = <factory>, sensor: ‘Go2ArmSensor’ = <factory>, ik: ‘IKConfig’ = <factory>, asset: ‘Asset’ = <factory>, iterations: ‘int | None’ = None, model_file: ‘str’ = <factory>, init_state: ‘InitState’ = <factory>, commands: ‘CommandsConfig’ = <factory>, reward_config: ‘RewardConfig | None’ = None, domain_rand: ‘Go2ArmDomainRandConfig’ = <factory>, goal_ee: ‘EEGoalConfig’ = <factory>, history: ‘HistoryConfig’ = <factory>, arm_stage: ‘ArmStageConfig’ = <factory>, curriculum: ‘CurriculumConfig’ = <factory>)

Parameters:

scene (SceneCfg)
sim_dt (float)
max_episode_seconds (float)
ctrl_dt (float)
render_spacing (float)
render_offset_mode (str)
motrix_max_iterations (Optional[int])
post_step_forward_sensor (bool)
control_config (ControlConfig)
noise_config (NoiseConfig)
sensor (Go2ArmSensor)
ik (IKConfig)
asset (Asset)
iterations (int | None)
model_file (str)
init_state (InitState)
commands (CommandsConfig)
reward_config (RewardConfig | None)
domain_rand (Go2ArmDomainRandConfig)
goal_ee (EEGoalConfig)
history (HistoryConfig)
arm_stage (ArmStageConfig)
curriculum (CurriculumConfig)

model_file: str¶

max_episode_seconds: float = 20.0¶

init_state: InitState¶

commands: CommandsConfig¶

reward_config: RewardConfig | None = None¶

sensor: Go2ArmSensor¶

domain_rand: Go2ArmDomainRandConfig¶

goal_ee: EEGoalConfig¶

history: HistoryConfig¶

arm_stage: ArmStageConfig¶

curriculum: CurriculumConfig¶

__init__(scene=<factory>, sim_dt=0.01, max_episode_seconds=20.0, ctrl_dt=0.02, render_spacing=1.0, render_offset_mode='grid', motrix_max_iterations=None, post_step_forward_sensor=False, control_config=<factory>, noise_config=<factory>, sensor=<factory>, ik=<factory>, asset=<factory>, iterations=None, model_file=<factory>, init_state=<factory>, commands=<factory>, reward_config=None, domain_rand=<factory>, goal_ee=<factory>, history=<factory>, arm_stage=<factory>, curriculum=<factory>)¶

Parameters:

scene (SceneCfg)
sim_dt (float)
max_episode_seconds (float)
ctrl_dt (float)
render_spacing (float)
render_offset_mode (str)
motrix_max_iterations (Optional[int])
post_step_forward_sensor (bool)
control_config (ControlConfig)
noise_config (NoiseConfig)
sensor (Go2ArmSensor)
ik (IKConfig)
asset (Asset)
iterations (int | None)
model_file (str)
init_state (InitState)
commands (CommandsConfig)
reward_config (RewardConfig | None)
domain_rand (Go2ArmDomainRandConfig)
goal_ee (EEGoalConfig)
history (HistoryConfig)
arm_stage (ArmStageConfig)
curriculum (CurriculumConfig)

class unilab.envs.locomotion.go2_arm.manip_loco.Go2ArmManipLocoDRProvider[source]¶

Bases: LocomotionDRProvider

Parameters:

base_kp (ndarray | None)
base_kd (ndarray | None)
base_body_mass (ndarray | None)
base_geom_friction (ndarray | None)
ground_geom_id (int | None)
base_dof_armature (ndarray | None)

__init__(*, base_kp=None, base_kd=None, base_body_mass=None, base_geom_friction=None, ground_geom_id=None, base_dof_armature=None)[source]¶

Parameters:

base_kp (ndarray | None)
base_kd (ndarray | None)
base_body_mass (ndarray | None)
base_geom_friction (ndarray | None)
ground_geom_id (int | None)
base_dof_armature (ndarray | None)

build_reset_plan(env, env_ids)[source]¶

Parameters:

env (Any)
env_ids (ndarray)

Return type:

ResetPlan

class unilab.envs.locomotion.go2_arm.manip_loco.Go2ArmManipLocoEnv[source]¶

Bases: Go2ArmBaseEnv

Parameters:: cfg (Go2ArmManipLocoCfg)

__init__(cfg, num_envs=1, backend_type='mujoco')[source]¶

Parameters:: cfg (Go2ArmManipLocoCfg)

property obs_groups_spec: dict[str, int]¶

101}.

Subclasses MUST override this property.

Type:: Return observation group dimensions, e.g. {“obs”
Type:: 98, “critic”

reset_ee_goals(env_ids)[source]¶

Reset EE goals by sampling the first segment from init_ee_cart.

Parameters:: env_ids (ndarray)
Return type:: None

apply_action(actions, state)[source]¶

子类实现：action → ctrl

Parameters:

actions (ndarray)
state (NpEnvState)

Return type:

ndarray

update_state(state)[source]¶

子类实现：计算 obs/reward/terminated

Parameters:: state (NpEnvState)
Return type:: NpEnvState