Term Catalog

Every term registered in Eden (246 of them) with the parameters it accepts, generated from the registries themselves. This is the complete list; the guide's Terms & Registries page explains how to write and wire one up.

• Bold parameters are required — omit one and configure() raises before the env is built.
• A term's parameters are the keys of its params={...} dict (function terms) or its configure() fields (class terms). The fields every term of a kind carries are listed once, under that kind.
• A Requires column appears in the tables that have one: an EnvOptions flag the term needs set, which no parameter of the term can supply. The env build raises without it.
• eden terms prints the same catalog in a terminal, eden terms --kind reward one section of it, eden terms --grep height only the terms whose name, parameters or summary match, and eden terms --json emits it as JSON.

action terms

14 registered.

• Class terms: <ClassName>.configure(<field>=...) — import it from the module its section names
• Every action term also accepts: dofs_name: str \| list[str] = [], entity_name: str = ''

eden.managers.terms.actions.binary_actions

Term	Form	Parameters	Summary
binary_joint_controller	class BinaryJointController	close_action: float \| dict[str, float] = 0.0, open_action: float \| dict[str, float] = 1.0, scale: float = 1.0	A binary joint controller for a parallel-jaw gripper.
gripper_decision_controller	class GripperDecisionController	close_action: float \| dict[str, float] = 0.0, close_threshold: float \| None = None, continuous: bool = False, hold_steps: int = 0, input: GripperInput = 'sign', on_missing: Literal['hold', 'measured'] = 'hold', open_action: float \| dict[str, float] = 1.0, open_threshold: float \| None = None, scale: float = 1.0, snap_threshold: float \| None = None, sticky_repeats: int = 0, threshold: float = 0.5, zero_holds: bool = False	Gripper controller that decides open/close from a policy scalar.

eden.managers.terms.actions.free_base_actions

Term	Form	Parameters	Summary
free_base_pose_controller	class FreeBasePoseController	force_limit: float = 100.0, orientation_kd: float = 2.0, orientation_kp: float = 40.0, orientation_scale: float \| tuple[float, float, float] = 0.2, position_kd: float = 60.0, position_kp: float = 1000.0, position_scale: float \| tuple[float, float, float] = 0.05, torque_limit: float = 10.0	Drive a floating base to a pose near its default root pose (Genesis free-joint PD).

eden.managers.terms.actions.joint_actions

Term	Form	Parameters	Summary
explicit_pd_controller	class ExplicitPDController	dict_keys_scope: Literal['term', 'entity'] = 'term', mimic_dofs: dict[str, tuple[str, float, float]] = {}, modifier: ActionModifierOptions \| None = None, offset: float \| dict[str, float] = 0.0, reference_source: ReferenceSource = ReferenceSource.DEFAULT, scale: float \| dict[str, float] \| None = None, scale_ratio: float \| None = None	Explicit joint PD controller using manual torque computation.
explicit_velocity_controller	class ExplicitVelocityController	dict_keys_scope: Literal['term', 'entity'] = 'term', mimic_dofs: dict[str, tuple[str, float, float]] = {}, modifier: ActionModifierOptions \| None = None, offset: float \| dict[str, float] = 0.0, reference_source: ReferenceSource = ReferenceSource.ZERO, scale: float \| dict[str, float] \| None = None, scale_ratio: float \| None = None, velocity_gain: float = 0.25	Explicit joint velocity controller using manual torque computation.
implicit_pd_controller	class ImplicitPDController	dict_keys_scope: Literal['term', 'entity'] = 'term', mimic_dofs: dict[str, tuple[str, float, float]] = {}, modifier: ActionModifierOptions \| None = None, offset: float \| dict[str, float] = 0.0, reference_source: ReferenceSource = ReferenceSource.DEFAULT, scale: float \| dict[str, float] \| None = None, scale_ratio: float \| None = None	Implicit joint PD controller using Genesis's built-in position control.
implicit_velocity_controller	class ImplicitVelocityController	dict_keys_scope: Literal['term', 'entity'] = 'term', mimic_dofs: dict[str, tuple[str, float, float]] = {}, modifier: ActionModifierOptions \| None = None, offset: float \| dict[str, float] = 0.0, reference_source: ReferenceSource = ReferenceSource.ZERO, scale: float \| dict[str, float] \| None = None, scale_ratio: float \| None = None	Joint velocity control through Genesis's built-in velocity mode.
null_joint_action	class NullJointAction	—	Virtually fix the DoFs to the default positions.
velocity_feedforward_pd_controller	class VelocityFeedforwardPDController	dict_keys_scope: Literal['term', 'entity'] = 'term', feed_forward_ratio: float = 1.0, mimic_dofs: dict[str, tuple[str, float, float]] = {}, modifier: ActionModifierOptions \| None = None, offset: float \| dict[str, float] = 0.0, reference_source: ReferenceSource = ReferenceSource.DEFAULT, scale: float \| dict[str, float] \| None = None, scale_ratio: float \| None = None, velocity_scale: float = 1.0	Explicit joint PD controller with velocity feedforward.

eden.managers.terms.actions.task_space_actions

Term	Form	Parameters	Summary
differential_ik_controller	class DifferentialIKController	damping: float = 0.1, ee_link_name: str = '', ee_local_pos: tuple[float, float, float] = (0.0, 0.0, 0.0), max_dq: float \| None = None, orientation_scale: float \| list[float] = 0.5, orientation_weight: float = 1.0, position_scale: float \| list[float] = 0.05, position_weight: float = 1.0	Action term that maps task-space commands to joint targets via differential IK.
ik_pose_controller	class IKPoseController	ee_link_name: str = 'gripper_link', ee_link_offset_pos: tuple[float, float, float] = (0.0, 0.0, 0.0), ee_link_offset_quat: tuple[float, float, float, float] = (1.0, 0.0, 0.0, 0.0), joint_limit_margin: float = 0.0, max_ik_position_error: float \| None = None, max_ik_rotation_error: float \| None = None, max_joint_delta: float \| None = None	Absolute task-space pose controller: solve IK for a base-frame target pose each step.
operational_space_controller	class OperationalSpaceController	action_clip: float \| tuple[float, ...] = 1.0, control_mode: Literal['position', 'pose'] = 'pose', damping_ratio: float \| tuple[float, ...] = 1.0, ee_link_name: str = '', ee_local_pos: tuple[float, float, float] = (0.0, 0.0, 0.0), ee_local_quat: tuple[float, float, float, float] = (1.0, 0.0, 0.0, 0.0), inertial_decoupling: Literal['none', 'full', 'position_orientation'] = 'none', kp: float \| tuple[float, ...] = 100.0, nullspace_damping_ratio: float = 1.0, nullspace_kp: float \| None = None, nullspace_reference: Literal['default', 'reset_pose'] = 'default', operational_inertia_damping: float = 1e-06, orientation_scale: float \| tuple[float, float, float] = 0.5, position_scale: float \| tuple[float, float, float] = 0.05	Operational-space control to a persistent relative position/pose target.

eden.managers.terms.actions.welding_actions

Term	Form	Parameters	Summary
parallel_jaw_welding	class ParallelJawWelding	ee_link_name: str = '', grasp_force_threshold: float = 1.0, ignore_entities_name: list[str] = [], ignore_links_idx: list[int] = [], left_gripper_link_name: str = '', right_gripper_link_name: str = ''	Weld the link a parallel-jaw gripper is squeezing to the end-effector link.
suction_cup_welding	class SuctionCupWelding	ee_link_name: str = '', ignore_entities_name: list[str] = [], ignore_links_idx: list[int] = []	Collision-aware welding of an object link to the suction-cup end-effector link.

observation terms

64 registered.

• Function terms: ObsTerm.configure(func=observations.<term>, params={...}) — from eden import observations, from eden.terms import ObsTerm
• Class terms: <ClassName>.configure(<field>=...) — import it from the module its section names
• Every observation term also accepts: backfill: bool = True, clip: tuple[float, float] \| None = None, flatten_history_dim: bool = True, history_length: int = 0, modifiers: list[eden.options.managers.modifiers.ObservationModifierOptions] = [], noise: eden.options.managers.modifiers.NoiseOptions \| None = None, scale: float \| list[float] = 1.0

eden.managers.terms.observations.camera

Term	Form	Parameters	Summary
depth_image	function	camera_name: str	—
normal_image	function	camera_name: str	—
pointcloud_image	function	camera_name: str, world_frame: bool = False	—
rgb_image	function	camera_name: str, normalize: bool = False, channels_first: bool = False	RGB image from a camera, optionally CNN-ready.
segmentation_image	function	camera_name: str	—

eden.managers.terms.observations.common

Term	Form	Parameters	Summary
episode_phase	function	centered: bool = False	Return normalized episode time as a (num_envs, 1) tensor.
generated_commands	function	command_name: str	—
last_action	function	action_name: str \| None = None	—
metric_observation	class MetricObservation	metric_name: str = ''	Expose a MetricTerm's cached value as a (num_envs, 1) observation.

eden.managers.terms.observations.contact

Term	Form	Parameters	Summary
contact_force_norm	class ContactForceNorm	entity_name: str = 'robot', links_name: list[str] = []	Per-link contact-force magnitude (Euclidean, world frame).
grouped_contact_flags	class GroupedContactFlags	entity_name: str = 'robot', force_threshold: float \| None = None, links_name: list[str] = [], object_entity_name: str = 'object', object_links_name: list[str] = []	Per-(link, partner-link) boolean contact flags with a specific entity.
motion_contact_flags	function	entity_name: str, links_name: list[str]	Demo per-(link, partner) contact flags at the current motion frame.
motion_future_contact_label	function	entity_name: str, frame_offsets: Sequence[int], links_name: list[str] \| None = None, flatten: bool = True	Demo tri-state per-link contact labels (+1 contact / -1 no contact / 0 unlabelled) at future frames.

eden.managers.terms.observations.locomotion

Term	Form	Parameters	Summary
base_height	function	entity_name: str = 'robot', terrain_entity_name: str \| None = None, relative: bool = False	Height of the entity's base above the ground, shape (num_envs, 1).
cos_phase	function	command_name: str	Cosine of the gait phase for each foot. Shape (num_envs, num_feet).
grid_scan_dots	class GridScanDots	clamp_offset_range: Vec2FType = (-5.0, 5.0), entity_name: str = 'robot', flatten_measured_points: bool = True, measured_points_x: list[float] = [-0.8, -0.7, -0.6, -0.5, -0.4, -0.3, -0.2, -0.1, 0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8], measured_points_y: list[float] = [-0.5, -0.4, -0.3, -0.2, -0.1, 0.0, 0.1, 0.2, 0.3, 0.4, 0.5], offset_from_base: bool = True, terrain_entity_name: str = 'terrain'	Height scan of the terrain at the given xy coordinates.
projected_gravity	class ProjectedGravity	entity_name: str = 'robot'	Gravity direction expressed in the entity's base frame.
scaled_commands	class ScaledCommands	command_name: str = '', command_scale: list[float] = []	Command vector multiplied by a per-component scale; prefer the built-in scale=[...] modifier.
sin_phase	function	command_name: str	Sine of the gait phase for each foot. Shape (num_envs, num_feet).
velocity_tracking_error_integral	class VelocityTrackingErrorIntegral	command_name: str = 'velocity_command', entity_name: str = 'robot', max_abs: float = 1.0	Running integral of the base-velocity tracking error (N, 3).

eden.managers.terms.observations.manipulation

Term	Form	Parameters	Summary
end_effector_pose	class EndEffectorPose	ee_link_name: str = '', entity_name: str = 'robot', frame: Literal['base', 'world'] = 'base', frame_correction_quat: tuple[float, float, float, float] = (1.0, 0.0, 0.0, 0.0), offset_pos: tuple[float, float, float] = (0.0, 0.0, 0.0), offset_quat: tuple[float, float, float, float] = (1.0, 0.0, 0.0, 0.0), rotation: RotationFormat = 'quat'	End-effector pose [pos(3), rotation(k)] of a link, in the base or world frame.
end_effector_to_object_distance	class EndEffectorToObjectDistance	entity_link_name: str \| None = None, entity_name: str = 'obj', frame: Literal['world', 'robot'] = 'robot', left_ee_link_name: str = 'left_gripper', right_ee_link_name: str = 'right_gripper', robot_name: str = 'robot'	End-effector-to-object vector (N, 3).
gripper_state	class GripperState	close_pos: float \| None = None, dofs_name: list[str] = [], entity_name: str = 'robot', normalize: GripperNormalize = 'raw', open_pos: float \| None = None	Gripper DOF positions as raw values, openness or closedness.
object_to_oracle_distance	class ObjectToOracleDistance	command_name: str = 'command', entity_link_name: str \| None = None, entity_name: str = 'obj', frame: Literal['world', 'robot'] = 'robot', robot_name: str = 'robot'	Object-to-oracle vector (N, 3).

eden.managers.terms.observations.privileged

Term	Form	Parameters	Summary
added_mass	class AddedMass	entity_name: str = 'robot', links_name: list[str] = []	Per-env mass shift (kg) of the named links vs nominal — 0 without mass DR.
com_offset	class ComOffset	entity_name: str = 'robot', links_name: list[str] = []	Per-env COM shift (m) of the named links vs nominal — 0 without COM DR.
foot_friction	class FootFriction	entity_name: str = 'robot', links_name: list[str] = []	Mean per-env friction ratio over the named links — 1.0 without friction DR.
gain_scale	class GainScale	entity_name: str = 'robot'	Per-DOF relative deviation of the realized PD gains from their defaults.
motor_strength_scale	class MotorStrengthScale	action_term_name: str = 'dofs_pos_controller'	Per-DOF deviation of the realized actuator torque-scale from its nominal.

eden.managers.terms.observations.proprio

Term	Form	Parameters	Summary
base_ang_vel	function	entity_name: str, frame: Literal['world', 'body'] = 'world'	—
base_lin_vel	function	entity_name: str, frame: Literal['world', 'body'] = 'world'	—
base_pos	function	entity_name: str, relative: bool = False	Return the position of the entity's base link.
base_quat	function	entity_name: str, relative: bool = False	Return the quaternion of the entity's base link.
base_rpy	function	entity_name: str, terms_idx: torch.Tensor \| None = None	Return the roll, pitch, and yaw of the entity's base link.
dofs_control_force	class DofsControlForce	dofs_name: list[str] \| None = None, entity_name: str = 'robot', terms_idx: list[int] \| None = None	Control force the actuators applied on each of an entity's DOFs over the last physics step
dofs_force	class DofsForce	dofs_name: list[str] \| None = None, entity_name: str = 'robot', terms_idx: list[int] \| None = None	Generalized force the solver applied to each of an entity's DOFs (get_dofs_force).
dofs_pos	class DofsPos	dofs_name: list[str] \| None = None, entity_name: str = 'robot', offset_from_default: bool = False, terms_idx: list[int] \| None = None	Joint positions (angles; positions for linear actuators) of an entity's DOFs.
dofs_vel	class DofsVel	dofs_name: list[str] \| None = None, entity_name: str = 'robot', terms_idx: list[int] \| None = None	Joint velocities of an entity's DOFs.
links_ang	function	entity_name: str, ls_idx_local: torch.Tensor \| None = None	—
links_pos	function	entity_name: str, ls_idx_local: torch.Tensor \| None = None	—
links_quat	function	entity_name: str, ls_idx_local: torch.Tensor \| None = None	—
links_vel	function	entity_name: str, ls_idx_local: torch.Tensor \| None = None	—

eden.managers.terms.observations.sensors

Term	Form	Parameters	Summary
sensor_read	class SensorRead	post_process_func: Callable \| None = None, read_ground_truth: bool = False, sensor_names: list[str] \| None = None	Concatenate selected sensor readings into a single observation tensor.

eden.managers.terms.observations.tracking

Term	Form	Parameters	Summary
entity_base_ori_b	function	entity_name: str, anchor_entity_name: str, anchor_link_name: str, representation: Literal['rot6d', 'tan_norm'] = 'rot6d'	Live base orientation of entity_name in another entity's live anchor-link frame, 6-D. Shape (N, 6).
entity_base_pos_b	function	entity_name: str, anchor_entity_name: str, anchor_link_name: str	Live base position of entity_name in another entity's live anchor-link frame. Shape (N, 3).
motion_base_pos_w	function	entity_name: str	World-frame motion-ref base position. Shape (num_envs, 3).
motion_base_quat_w	function	entity_name: str	World-frame motion-ref base quaternion (wxyz). Shape (num_envs, 4).
motion_dofs_pos	function	entity_name: str, dofs_name: list[str] \| None = None	Motion-ref qpos. Shape (num_envs, D).
motion_dofs_vel	function	entity_name: str, dofs_name: list[str] \| None = None	Motion-ref qvel. Symmetric to motion_dofs_pos.
motion_future_base_ori_b	function	entity_name: str, anchor_entity_name: str, anchor_link_name: str, frame_offsets: Sequence[int], representation: Literal['rot6d', 'tan_norm'] = 'rot6d', flatten: bool = True	Future-frame motion-ref base orientation of entity_name in another entity's live anchor-link frame.
motion_future_base_pos_b	function	entity_name: str, anchor_entity_name: str, anchor_link_name: str, frame_offsets: Sequence[int], flatten: bool = True	Future-frame motion-ref base position of entity_name in another entity's live anchor-link frame.
motion_future_base_pos_w	function	entity_name: str, frame_offsets: Sequence[int], flatten: bool = True	World-frame motion-ref base position at multiple buffer frames.
motion_future_base_quat_w	function	entity_name: str, frame_offsets: Sequence[int], flatten: bool = True	World-frame motion-ref base quaternion (wxyz) at multiple buffer frames.
motion_future_dofs_pos	function	entity_name: str, frame_offsets: Sequence[int], dofs_name: list[str] \| None = None, flatten: bool = True	Return motion-ref qpos at multiple future buffer frames.
motion_future_dofs_vel	function	entity_name: str, frame_offsets: Sequence[int], dofs_name: list[str] \| None = None, flatten: bool = True	Return motion-ref qvel at multiple future buffer frames.
motion_future_links_ori_b	function	entity_name: str, links_name: list[str], anchor_link_name: str, frame_offsets: Sequence[int], representation: Literal['rot6d', 'tan_norm'] = 'rot6d', flatten: bool = True	Return future-frame motion-ref link orientations in the robot's current anchor frame.
motion_future_links_pos_b	function	entity_name: str, links_name: list[str], anchor_link_name: str, frame_offsets: Sequence[int], flatten: bool = True	Return future-frame motion-ref links_name positions in the robot's current anchor frame.
motion_links_ori_b	function	entity_name: str, links_name: list[str], anchor_link_name: str, flatten: bool = True	Return motion-ref links_name orientations in the robot's anchor_link_name frame.
motion_links_pos_b	function	entity_name: str, links_name: list[str], anchor_link_name: str, flatten: bool = True	Return motion-ref links_name positions in the robot's anchor_link_name frame.
motion_links_pos_w	function	entity_name: str, links_name: list[str], flatten: bool = True	Return world-frame motion-ref link positions.
motion_links_quat_w	function	entity_name: str, links_name: list[str], flatten: bool = True	Return world-frame motion-ref link quaternions (wxyz).
motion_phase	function	entity_name: str	Normalized clip playback phase in [0, 1]. Shape (num_envs, 1).
robot_body_ori_b	function	entity_name: str, anchor_link_name: str, tracked_link_names: list[str], flatten: bool = True	Return robot body orientations relative to its own anchor, as 2-column rotation matrices.
robot_body_pos_b	function	entity_name: str, anchor_link_name: str, tracked_link_names: list[str], flatten: bool = True	Return robot body positions relative to its own anchor link.

reward terms

46 registered.

• Function terms: RewardTerm.configure(func=rewards.<term>, params={...}) — from eden import rewards, from eden.terms import RewardTerm
• Class terms: <ClassName>.configure(<field>=...) — import it from the module its section names
• Every reward term also accepts: range_s: list[tuple[float, float]] \| None = None, tags: list[str] = [], weight: float = 1.0

eden.managers.terms.rewards.common

Term	Form	Parameters	Summary
action_l2	function	action_term_name: str \| None = None	Penalize the actions using L2 squared kernel.
action_rate_l2	function	action_term_name: str \| None = None	Penalize the rate of change of the actions using L2 squared kernel.
action_rate_l2_smooth	class ActionRateL2Smooth	action_term_name: str \| None = None	Penalize the 1st and 2nd derivatives of the actions using an L2 squared kernel.
dofs_acc_l2	class DofsAccL2	entity_name: str = 'robot'	Penalize joint accelerations approximated as velocity differences.
dofs_pos_limits	function	entity_name: str	Penalize joint positions if they cross the soft limits.
dofs_torques	function	entity_name: str	Penalize joint torques: sqrt(sum(tau^2)) + sum(|tau|).
dofs_vel_limits	function	entity_name: str, max_vel: float \| torch.Tensor	Quadratic hinge penalty on joint velocities exceeding a symmetric limit.
metric_reward	class MetricReward	allow_cumulative: bool = False, metric_name: str = ''	Use a MetricTerm's cached value as the reward.
self_collision_cost	function	entity_name: str	Penalize self collisions.
similar_to_default	function	entity_name: str	Penalize joint poses far away from the entity's default pose.
task_success	class TaskSuccess	—	1 on a step where every success metric holds.
undesired_contacts	class UndesiredContacts	entity_name: str = 'robot', links_name: list[str] = [], threshold: float = 1.0	Penalty count of links whose net contact-force magnitude exceeds threshold.

eden.managers.terms.rewards.contact

Term	Form	Parameters	Summary
motion_contact_chamfer_reward	class MotionContactChamferReward	empty_distance: float = 100.0, frame: Literal['world', 'object', 'wrist'] = 'world', hand_entity_name: str = 'robot', hand_links_name: list[str] = [], kernel: Literal['exp', 'sigmoid'] = 'exp', lookahead_index: int = 0, object_entity_name: str = 'object', object_links_name: list[str] = [], sigmoid_a: float = 100.0, sigmoid_b: float = 5.0, std: float = 0.18257418583505536, wrist_link_name: str \| None = None, zero_contact_mode: Literal['zero', 'one'] = 'zero'	Exp/sigmoid reward on the policy-vs-demo contact Chamfer distance.
motion_contact_distance_reward	class MotionContactDistanceReward	hand_entity_name: str = 'robot', hand_links_name: list[str] = [], object_entity_name: str = 'object', object_links_name: list[str] = [], std: float = 0.03	Exp-kernel reward on per-link contact-point distance in the object frame.
motion_contact_match_reward	class MotionContactMatchReward	hand_entity_name: str = 'robot', hand_links_name: list[str] = [], object_entity_name: str = 'object', object_links_name: list[str] = []	Fraction of links whose contact state matches the demo's contact-ON links.

eden.managers.terms.rewards.locomotion

Term	Form	Parameters	Summary
base_height_l2	function	entity_name: str, target_height: float, command_name: str \| None = None, zero_vel_penalty_scale: float = 1.0, zero_vel_threshold: float = 0.1, terrain_entity_name: str \| None = None, relative: bool = False	L2 base-height penalty with optional zero-velocity amplification.
body_ang_vel_xy_penalty	class BodyAngVelXYPenalty	entity_name: str = 'robot', link_name: str = ''	Penalize body angular velocity in xy (allow z rotation).
close_feet_penalty	class CloseFeetPenalty	entity_name: str = 'robot', feet_link_names: list[str] = [], threshold: float = 0.15	Penalize when feet are too close together in the xy plane.
close_feet_yaw_projected_penalty	class CloseFeetYawProjectedPenalty	entity_name: str = 'robot', feet_link_names: list[str] = [], threshold: float = 0.05	Penalize feet closer than a lateral distance in the base's yaw frame.
constant_base_height	function	entity_name: str, target_height: float, terrain_entity_name: str \| None = None, relative: bool = False	Penalize base height away from target — base_height_l2 without velocity gating.
feet_air_time	class FeetAirTime	command_name: str = '', command_threshold: float = 0.1, contact_force_threshold: float = 2.0, entity_name: str = 'robot', feet_link_names: list[str] = [], target_s: float = 0.35	Legged-gym style feet air-time reward for aperiodic gaits.
feet_clearance	class FeetClearance	command_name: str = '', command_threshold: float = 0.1, entity_name: str = 'robot', feet_link_names: list[str] = [], target_height: float = 0.0	Penalize deviation from target clearance height, weighted by foot xy velocity.
feet_contact_balance_penalty	class FeetContactBalancePenalty	contact_force_threshold: float = 2.0, ema_alpha: float = 0.02, entity_name: str = 'robot', feet_link_names: list[str] = []	Penalize uneven foot usage (a limp) via the variance of per-foot contact duty.
feet_orientation_penalty	class FeetOrientationPenalty	entity_name: str = 'robot', feet_link_names: list[str] = []	Penalize feet orientation deviation from flat (projected gravity in foot frame).
feet_phase	class FeetPhase	entity_name: str = 'robot', feet_link_names: list[str] = [], gait_command_name: str = '', swing_height: float = 0.08, tracking_sigma: float = 0.25	Reward for tracking desired foot height based on gait phase.
feet_slip	class FeetSlip	command_name: str = '', command_threshold: float = 0.1, contact_height_threshold: float = 0.05, entity_name: str = 'robot', feet_link_names: list[str] = []	Penalize foot sliding (xy velocity while in contact with ground).
flat_orientation_l2	class FlatOrientationL2	entity_name: str = 'robot'	Penalize non-flat base orientation (projected gravity deviation in xy).
is_terminated_penalty	function	—	Return 1 for terminated environments.
lin_vel_z	function	entity_name: str	Penalize z axis base linear velocity.
stumble_penalty	class StumblePenalty	entity_name: str = 'robot', feet_link_names: list[str] = [], horizontal_to_vertical_ratio: float = 4.0	Compute an indicator penalty for feet hitting vertical surfaces (e.g. stairs riser, walls).
track_ang_vel_with_xy_penalty	function	command_name: str, entity_name: str, std: float	Track commanded angular velocity with additional xy angular velocity penalty.
track_lin_vel_with_z_penalty	function	command_name: str, entity_name: str, std: float	Track commanded linear velocity with additional z-velocity penalty.
tracking_ang_vel	class TrackingAngVel	command_name: str = '', entity_name: str = 'robot', sigma: float = 0.25	Tracking of angular velocity commands (yaw).
tracking_lin_vel	class TrackingLinVel	command_name: str = '', entity_name: str = 'robot', sigma: float = 0.25	Tracking of linear velocity commands (xy axes).
upright	class Upright	entity_name: str = 'robot', link_name: str = '', std: float = 1.0	Reward for keeping a body link upright.
variable_posture	class VariablePosture	command_name: str = '', entity_name: str = 'robot', running_threshold: float = 1.5, std_running: dict[str, float] = {}, std_standing: dict[str, float] = {}, std_walking: dict[str, float] = {}, walking_threshold: float = 0.1	Speed-dependent posture reward.
weighted_pose_penalty	class WeightedPosePenalty	entity_name: str = 'robot', pose_weights: list[float] = []	Penalize deviation from default pose with per-joint weights.

eden.managers.terms.rewards.tracking

Term	Form	Parameters	Summary
motion_base_orientation_tracking_exp	function	entity_name: str, std: float	Exp-kernel reward on quat-error magnitude of the base vs motion ref (wxyz).
motion_base_position_tracking_exp	function	entity_name: str, std: float	Exp-kernel reward on world-frame base position vs motion ref.
motion_dofs_pos_tracking_exp	function	entity_name: str, std: float, dofs_name: list[str] \| None = None, reduction: Literal['sum', 'per_dof_mean', 'mean'] = 'sum', frame_offsets: Sequence[int] = (0,)	Exp-kernel reward on per-DOF qpos vs motion ref.
motion_global_body_angular_velocity_error_exp	class MotionGlobalBodyAngularVelocityErrorExp	entity_name: str = 'robot', frame_offsets: tuple[int, ...] = (0,), links_name: list[str] = [], std: float = 0.1	—
motion_global_body_linear_velocity_error_exp	class MotionGlobalBodyLinearVelocityErrorExp	entity_name: str = 'robot', frame_offsets: tuple[int, ...] = (0,), links_name: list[str] = [], std: float = 0.1	—
motion_link_orientation_tracking_exp	function	entity_name: str, links_name: list[str], std: float, exp_first: bool = True, frame_offsets: Sequence[int] = (0,)	Exp-kernel reward on world-frame link orientations vs motion ref (wxyz).
motion_link_position_tracking_exp	function	entity_name: str, links_name: list[str], std: float, exp_first: bool = True, frame_offsets: Sequence[int] = (0,)	Exp-kernel reward on world-frame link positions vs motion ref.
motion_relative_body_orientation_error_exp	class MotionRelativeBodyOrientationErrorExp	anchor_link_name: str = 'torso_link', entity_name: str = 'robot', frame_offsets: tuple[int, ...] = (0,), links_name: list[str] = [], std: float = 0.1	Yaw-warped multi-link orientation tracking (exp kernel, link mean).
motion_relative_body_position_error_exp	class MotionRelativeBodyPositionErrorExp	anchor_link_name: str = 'torso_link', entity_name: str = 'robot', frame_offsets: tuple[int, ...] = (0,), links_name: list[str] = [], std: float = 0.1	Yaw-warped multi-link position tracking (exp kernel, link mean).

termination terms

34 registered.

• Function terms: DoneTerm.configure(func=terminations.<term>, params={...}) — from eden import terminations, from eden.terms import DoneTerm
• Class terms: <ClassName>.configure(<field>=...) — import it from the module its section names
• Every termination term also accepts: time_out: bool = False

eden.managers.terms.metrics.articulation

Term	Form	Parameters	Summary
joint_angle_at_target	function	entity_name: str, joint_name: str, target_angle: float, threshold: float = 0.05	Check if a single DOF is within threshold of target_angle.

eden.managers.terms.metrics.containment

Term	Form	Parameters	Summary
object_above_height	function	entity_name: str, height: float, relative: bool = False	Check if an entity's root z-position is above height.
object_in_region	function	entity_name: str, region_min: list[float], region_max: list[float], relative: bool = False	Check if an entity's root position is inside an axis-aligned box.

eden.managers.terms.metrics.grasp

Term	Form	Parameters	Summary
object_lifted	function	entity_name: str, min_height: float	Check if an entity has been lifted above min_height.

eden.managers.terms.metrics.proximity

Term	Form	Parameters	Summary
entity_near_entity_xy	function	entity_a_name: str, entity_b_name: str, threshold: float = 0.05	Check if two entities are within threshold horizontal (XY) distance.
entity_near_target	function	entity_name: str, target_pos: Vec3FType \| torch.Tensor, threshold: float = 0.05	Check if an entity is within threshold of a fixed world position.
object_a_is_on_b	function	entity_a_name: str, entity_b_name: str, xy_threshold: float = 0.03, height_threshold: float = 0.04, height_diff: float = 0.0	Check if object_a is on object_b by comparing height and xy distance of object origins.

eden.managers.terms.terminations.common

Term	Form	Parameters	Summary
cumulative_reward_below_threshold	function	reward_term: str, per_step_threshold: float, grace_steps: int = 0, check_interval: int = 1	Terminate envs whose running raw reward for reward_term falls below a threshold.
dofs_pos_limit_exceeded	function	entity_name: str, probability: float = 1.0	Terminate when any DOF position is outside the entity's soft position limits.
dofs_torque_limit_exceeded	function	entity_name: str, soft_ratio: float = 1.0, probability: float = 1.0	Terminate when any applied DOF torque exceeds soft_ratio * default_dofs_force_limits.
dofs_vel_limit_exceeded	function	entity_name: str, max_vel: float, probability: float = 1.0	Terminate when any DOF velocity exceeds max_vel in magnitude.
illegal_contact	function	entity_name: str, illegal_contact_entity_names: list[str]	—
illegal_link_contact	class IllegalLinkContact	entity_name: str = 'robot', force_threshold: float = 1.0, links_name: list[str] = []	Terminate when any of the named links carries a net contact force above force_threshold.
metric_above	class MetricAbove	metric_name: str = '', threshold: float = 0.0	Terminate the envs in which the named metric's cached value is strictly above threshold.
on_failure	function	—	—
on_metric_success	class OnMetricSuccess	metric_name: str = ''	Terminate the envs in which the named metric term reports success this step.
on_success	function	—	—
time_out	function	—	—

eden.managers.terms.terminations.locomotion

Term	Form	Parameters	Summary
base_orientation_pitch_greater_than	function	entity_name: str, threshold: float	Terminate when the base pitch exceeds threshold degrees.
base_orientation_roll_greater_than	function	entity_name: str, threshold: float	Terminate when the base roll exceeds threshold degrees.
root_height_below_minimum	function	entity_name: str, minimum_height: float, terrain_entity_name: str \| None = None, relative: bool = False	Terminate when the entity's base sinks below minimum_height above the ground.

eden.managers.terms.terminations.manipulation

Term	Form	Parameters	Summary
object_out_of_reach	function	entity_name: str, object_entity_name: str, object_link_name: str \| None = None, max_dist: float	Terminate when the base is over max_dist from the object in xy.

eden.managers.terms.terminations.tracking

Term	Form	Parameters	Summary
bad_anchor_ori	function	entity_name: str = 'robot', anchor_link_name: str = 'torso_link', threshold: float = 0.01	Terminate when the anchor's projected-gravity z diverges from the motion ref's.
bad_anchor_pos_z_only	function	entity_name: str = 'robot', anchor_link_name: str = 'torso_link', threshold: float = 0.01, threshold_adaptive: bool = False, down_threshold: float = 0.0, root_height_threshold: float = 0.5	Terminate when anchor link's |z| error vs motion ref exceeds threshold.
bad_base_ori	function	entity_name: str, threshold: float	Terminate when a free base's projected-gravity z diverges from the motion ref's by more than threshold.
bad_motion_link_pos_z_only	class BadMotionLinkPosZOnly	anchor_link_name: str = 'torso_link', down_threshold: float = 0.0, entity_name: str = 'robot', links_name: list[str] = [], root_height_threshold: float = 0.5, threshold: float = 0.01, threshold_adaptive: bool = False	Terminate when any tracked link's |z| error vs the motion ref exceeds threshold.
motion_base_orientation_tracking_error_above	function	entity_name: str, threshold: float	Terminate when base quat-error magnitude (wxyz) exceeds threshold.
motion_base_pose_tracking_break	class MotionBasePoseTrackingBreak	entity_name: str = 'object', pos_threshold: float = 0.1, rot_threshold: float = 1.0	Base-pose deviation gate vs the current motion frame.
motion_base_position_tracking_error_above	function	entity_name: str, threshold: float	Terminate when world-frame base-position L2 error exceeds threshold.
motion_dofs_pos_tracking_error_above	function	entity_name: str, threshold: float, dofs_name: list[str] \| None = None, reduction: Literal['mean', 'max'] = 'max'	Terminate when per-DOF |q − q_ref| reduction exceeds threshold.
motion_ends	function	entity_name: str	Terminate when the motion buffer's current lookahead frame is invalid.
motion_link_orientation_tracking_error_above	function	entity_name: str, links_name: list[str], threshold: float, reduction: Literal['mean', 'max'] = 'max'	Terminate when link quat-error reduction exceeds threshold.
motion_link_position_tracking_error_above	function	entity_name: str, links_name: list[str], threshold: float, reduction: Literal['mean', 'max'] = 'max'	Terminate when link-position L2 error reduction exceeds threshold.

eden.managers.terms.terminations.velocity_tracking

Term	Form	Parameters	Summary
fell_over	function	entity_name: str, limit_angle: float	Terminate when the robot leans too far from vertical.

command terms

7 registered.

• Class terms: <ClassName>.configure(<field>=...) — import it from the module its section names
• Every command term also accepts: debug_vis: bool = False, resampling_time_range: tuple[float, float] = None

eden.managers.terms.commands.gait

Term	Form	Parameters	Summary
gait_phase_command	class GaitPhaseCommand	gait_frequency: float = 1.25, num_feet: int = 2, phase_offset: float = 3.141592653589793, stand_phase_value: float = 3.141592653589793, stand_velocity_threshold: float = 0.01, velocity_command_name: str \| None = None	Command generator that maintains a cyclic gait phase for each foot.

eden.managers.terms.commands.mimic

Term	Form	Parameters	Summary
motion_command	class MotionCommand	adaptive_alpha: float = 0.001, adaptive_failure_rate_max_over_mean: float = 10.0, adaptive_kernel_size: int = 1, adaptive_lambda: float = 0.8, adaptive_uniform_ratio: float = 0.1, anchor_link_name: str = 'torso_link', base_joint_name: str \| None = None, base_link_name: str \| None = None, dofs_position_range: tuple[float, float] = (-0.1, 0.1), entity_name: str = '', ghost_entity_name: str = '', links_name: list[str] = ['pelvis', 'left_hip_roll_link', 'left_knee_link', 'left_ankle_roll_link', 'right_hip_roll_link', 'right_knee_link', 'right_ankle_roll_link', 'torso_link', 'left_shoulder_roll_link', 'left_elbow_link', 'left_wrist_yaw_link', 'right_shoulder_roll_link', 'right_elbow_link', 'right_wrist_yaw_link'], motion_file: str = 'motion.npz', motion_files: list[str] = [], pose_range: dict[str, tuple[float, float]] = {}, sampling_mode: Literal['adaptive', 'uniform', 'start'] = 'adaptive', velocity_range: dict[str, tuple[float, float]] = {}	Motion command for mimicking one or more motion clips.

eden.managers.terms.commands.oracle

Term	Form	Parameters	Summary
uniform_se3_command	class UniformSE3Command	pitch_range: tuple[float, float] = (-3.141592653589793, 3.141592653589793), roll_range: tuple[float, float] = (-3.141592653589793, 3.141592653589793), x_range: tuple[float, float] = (0.3, 0.5), y_range: tuple[float, float] = (-0.2, 0.2), yaw_range: tuple[float, float] = (-3.141592653589793, 3.141592653589793), z_range: tuple[float, float] = (0.2, 0.4)	Command that samples SE(3) translation and rotation command.
uniform_t3_command	class UniformT3Command	marker_color: tuple[float, float, float, float] = (0.1, 0.9, 0.1, 0.85), marker_entity_name: str = '', marker_frame_entity: str = '', marker_radius: float = 0.03, x_range: tuple[float, float] = (0.3, 0.5), y_range: tuple[float, float] = (-0.2, 0.2), z_range: tuple[float, float] = (0.2, 0.4)	Command that samples T(3) translation command.

eden.managers.terms.commands.velocity

Term	Form	Parameters	Summary
constant_velocity_command	class ConstantVelocityCommand	ang_vel_z_range: tuple[float, float] = (-1.0, 1.0), command_value: tuple[float, float, float] = (0.0, 0.0, 0.0), entity_name: str = '', lin_vel_x_range: tuple[float, float] = (-1.0, 1.0), lin_vel_y_range: tuple[float, float] = (-1.0, 1.0)	SE(2) velocity command held at a fixed value instead of resampled.
operator_velocity_command	class OperatorVelocityCommand	ang_vel_z_range: tuple[float, float] = (-1.0, 1.0), command_value: tuple[float, float, float] = (0.0, 0.0, 0.0), device_name: str = 'keyboard', device_options: dict = {}, entity_name: str = '', lin_vel_x_range: tuple[float, float] = (-1.0, 1.0), lin_vel_y_range: tuple[float, float] = (-1.0, 1.0)	SE(2) velocity command driven live by a human, from a keyboard or a gamepad.
uniform_velocity_command	class UniformVelocityCommand	ang_vel_z_range: tuple[float, float] = (0.0, 0.0), entity_name: str = '', heading_command: bool = False, heading_control_stiffness: float = 1.0, heading_range: tuple[float, float] \| None = None, lin_vel_x_range: tuple[float, float] = (0.0, 0.0), lin_vel_y_range: tuple[float, float] = (0.0, 0.0), rel_heading_envs: float = 1.0, rel_standing_envs: float = 0.0	SE(2) velocity command sampled from a uniform distribution.

event terms

32 registered.

• Function terms: EventTerm.configure(func=events.<term>, params={...}) — from eden import events, from eden.terms import EventTerm
• Class terms: <ClassName>.configure(<field>=...) — import it from the module its section names
• Every event term also accepts: interval_range_s: tuple[float, float] \| None = None, is_global_time: bool = False, min_step_count_between_reset: int = 0, mode: EventMode = EventMode.RESET, priority: int = 0

eden.managers.terms.events.domain_rand

Term	Form	Parameters	Requires	Summary
apply_external_force	class ApplyExternalForce	entity_name: str = 'robot', force_x_range: Vec2FType = (0.0, 0.0), force_y_range: Vec2FType = (0.0, 0.0), force_z_range: Vec2FType = (0.0, 0.0), links_name: list[str] = [], local: bool = False, prob: float = 1.0, ref: str = 'link_origin'	—	Apply a randomized external force on specified links.
apply_external_torque	class ApplyExternalTorque	entity_name: str = 'robot', links_name: list[str] = [], local: bool = False, prob: float = 1.0, ref: str = 'link_origin', torque_x_range: Vec2FType = (0.0, 0.0), torque_y_range: Vec2FType = (0.0, 0.0), torque_z_range: Vec2FType = (0.0, 0.0)	—	Apply a randomized external torque on specified links.
erfi50	class ERFI50	action_term_name: str = '', erfi_ratio: float = 0.5, lower_bound: float = -0.7777777777777778, upper_bound: float = 0.7777777777777778	—	ERFI-50 noise: 50% Random Actuation Offset (RAO) + 50% Random Force Injection (RFI).
push_by_velocity	class PushByVelocity	base_joint_name: str \| None = None, entity_name: str = 'robot', velocity_range: dict[str, Vec2FType] = {}	—	Periodically add a random velocity to a free base.
push_set_planar_velocity	class PushSetPlanarVelocity	base_joint_name: str \| None = None, entity_name: str = 'robot', velocity_xy_range: Vec2FType = (-0.5, 0.5)	—	Periodically SET the base planar velocity to a random value.
randomize_com_shift	class RandomizeComShift	com_shift_x_range: Vec2FType = (-0.01, 0.01), com_shift_y_range: Vec2FType = (-0.01, 0.01), com_shift_z_range: Vec2FType = (-0.01, 0.01), entity_name: str = 'robot', links_name: list[str] = []	EnvOptions(batch_links_info=True)	Offset the named links' centres of mass, per env.
randomize_constant_torque_kick	class RandomizeConstantTorqueKick	action_term_name: str = '', apply_as_ratio: bool = False, entity_name: str = 'robot', torque_kick_range: Vec2FType = (0.0, 0.01)	—	Randomize the torque magnitude of a ConstantTorqueKick action modifier.
randomize_deadband_epsilon	class RandomizeDeadbandEpsilon	action_term_name: str = '', apply_as_ratio: bool = False, deadband_epsilon_range: Vec2FType = (0.0, 0.01), entity_name: str = 'robot'	—	Randomize the position-error threshold of a Deadband action modifier.
randomize_dofs_pos_offset	class RandomizeDofsPosOffset	action_term_name: str \| list[str] \| None = None, dofs_pos_range: Vec2FType = (-0.01745, 0.01745), entity_name: str = ''	—	Per-env joint-position offset on joint position controllers (encoder bias).
randomize_friction_ratio	class RandomizeFrictionRatio	entity_name: str = 'robot', friction_range: Vec2FType = (0.3, 1.2), links_name: list[str] = []	—	Randomize sliding friction on links_name to absolute values, not a scale.
randomize_gear_backlash	class RandomizeGearBacklash	action_term_name: str = '', apply_as_ratio: bool = False, backlash_range: Vec2FType = (0.0, 0.01), entity_name: str = 'robot'	—	Randomize the one-sided target offset of a GearBacklash action modifier.
randomize_kp_kd_gains	class RandomizeKpKdGains	action_term_name: str \| None = None, entity_name: str = 'robot', kd_range: Vec2FType = (0.9, 1.1), kp_range: Vec2FType = (0.9, 1.1)	EnvOptions(batch_dofs_info=True)	Scale an entity's joint PD gains by a random factor.
randomize_link_mass_scale	class RandomizeLinkMassScale	entity_name: str = 'robot', links_name: list[str] = [], mass_scale_range: Vec2FType = (0.9, 1.1)	EnvOptions(batch_links_info=True)	Multiplicative per-link mass randomization.
randomize_mass_shift	class RandomizeMassShift	entity_name: str = 'robot', links_name: list[str] = [], mass_shift_range: Vec2FType = (-1.0, 3.0)	EnvOptions(batch_links_info=True)	Add a per-env offset to the named links' masses.
randomize_motor_strength	class RandomizeMotorStrength	action_term_name: str = '', entity_name: str = 'robot', motor_strength_range: Vec2FType = (0.9, 1.1)	—	Randomize a MotorStrength action modifier's scalar (default factor 0.9 to 1.1).
randomize_startup_dofs_pos	class RandomizeStartupDofsPos	apply_as_ratio: bool = False, dofs_pos_range: Vec2FType = (-0.01745, 0.01745), entity_name: str = 'robot'	—	Randomize the absolute joint positions of an entity's DOFs (default mode: reset).
randomize_startup_dofs_pos_bias	class RandomizeStartupDofsPosBias	action_terms_name: list[str] = [], dofs_pos_range: Vec2FType = (-0.01745, 0.01745), entity_name: str = 'robot'	—	Bake a per-env joint-zero bias into entity.default_dofs_pos at startup.
randomize_torque_noise	class RandomizeTorqueNoise	action_term_name: str = '', rfi_scale: float = 0.5	—	Residual Force Injection (RFI) — adds random torque noise to actuated joints.
set_random_dofs_pos	class SetRandomDofsPos	apply_as_ratio: bool = True, dofs_pos_range: Vec2FType = (0.0, 0.2), entity_name: str = 'robot'	—	Set the joint positions of an entity's DOFs to a random position (default mode: reset).

eden.managers.terms.events.motion

Term	Form	Parameters	Summary
object_reference_state_initialization	class ObjectReferenceStateInitialization	depenetration_window_s: float = 0.1, entity_name: str = 'object', max_depenetration_velocity: float \| None = None, xy_range: float = 0.0	On reset, set a free rigid object from the motion reference.
reference_state_initialization	class ReferenceStateInitialization	base_joint_name: str \| None = None, depenetration_window_s: float = 0.1, dofs_position_range: tuple[float, float] = (-0.1, 0.1), entity_name: str = 'robot', max_depenetration_velocity: float \| None = None, overall_noise_scale: float = 1.0, pose_range: dict[str, tuple[float, float]] = {}, velocity_range: dict[str, tuple[float, float]] = {}	On reset, initialise robot state from the motion reference (RSI).

eden.managers.terms.events.placement

Term	Form	Parameters	Summary
place_constant_xy	function	entity_name: str, support_entity_name: str, pos_xy: tuple[float, float], frame: Literal['world', 'local'] = 'world', support_idx: int \| None = None	—
place_from_layouts	class PlaceFromLayouts	layouts: list[dict[str, tuple[float, float, float]]] = [], support_entity_name: str = 'plane'	Each reset env draws one of a fixed list of layouts and places every entity the layout names.
place_gaussian	class PlaceGaussian	entity_name: str = '', mu_x: float = 0.0, mu_y: float = 0.0, support_entity_name: str = '', var_x: float = 0.1, var_y: float = 0.1	Event term that places an entity on a support surface via Gaussian-weighted sampling.
place_range_xy	function	entity_name: str, support_entity_name: str, range_x: tuple[float, float], range_y: tuple[float, float], range_roll: tuple[float, float] = (0, 0), range_pitch: tuple[float, float] = (0, 0), range_yaw: tuple[float, float] = (0, 0), frame: Literal['world', 'local'] = 'world', support_idx: int \| None = None	—
place_uniform	function	entity_name: str, support_entity_name: str, samples_per_area: int = 100	—

eden.managers.terms.events.resets

Term	Form	Parameters	Summary
reset_base_state_uniform	function	entity_name: str, pos_x_range: Vec2FType = (-0.5, 0.5), pos_y_range: Vec2FType = (-0.5, 0.5), pos_z_range: Vec2FType = (0.0, 0.0), roll_range: Vec2FType = (0.0, 0.0), pitch_range: Vec2FType = (0.0, 0.0), yaw_range: Vec2FType = (-3.14, 3.14)	Reset the base to its default pose, offset and rotated at random.
reset_dofs_by_offset	function	entity_name: str, dofs_pos_range: Vec2FType = (-0.1, 0.1), dofs_vel_range: Vec2FType \| None = None	Reset DOFs to default + U(offset), clamped to soft limits.
reset_dofs_by_scale	function	entity_name: str, dofs_pos_scale_range: Vec2FType = (0.95, 1.05)	Reset DOFs to default * U(scale) per env/DOF; velocities to 0.
reset_dofs_by_scale_and_offset	function	entity_name: str, dofs_scale_range: Vec2FType = (0.8, 1.2), dofs_offset_range: Vec2FType = (-0.1, 0.1)	Reset DOFs to default * U(scale) + U(offset); velocities to 0.

eden.managers.terms.events.solver

Term	Form	Parameters	Summary
set_contact_sol_params	function	sol_params: tuple[float, ...], entity_links: dict[str, list[str] \| None]	Set constraint-solver (solref/solimp) parameters on selected entity geoms.

eden.managers.terms.events.state_buffer

Term	Form	Parameters	Summary
reset_from_state_buffer	class ResetFromStateBuffer	category_weights: dict[str, float] \| None = None, path: str \| None = None, sampler: str \| None = None	Restore each resetting env to one stored row of a reset-state buffer file.

curriculum terms

6 registered.

• Class terms: <ClassName>.configure(<field>=...) — import it from the module its section names

eden.managers.terms.curricula.locomotion

Term	Form	Parameters	Summary
average_episode_length_tracker	class AverageEpisodeLengthTracker	num_compute_average_epl: int = 1000	Track the running mean of episode length.

eden.managers.terms.curricula.motion

Term	Form	Parameters	Summary
motion_base_pose_threshold_curriculum	class MotionBasePoseThresholdCurriculum	entity_name: str = 'object', pos_threshold_max: float = 0.1, pos_threshold_min: float = 0.05, rot_threshold_max: float = 1.0, rot_threshold_min: float = 0.5, success_ema_alpha: float = 0.05, success_rate_high: float = 0.999, success_rate_low: float = 0.9, termination_term_name: str = 'object_pose_break'	Tighten a MotionBasePoseTrackingBreak gate as success rises.

eden.managers.terms.curricula.rewards

Term	Form	Parameters	Summary
penalty_curriculum	class PenaltyCurriculum	degree: float = 0.00025, initial_scale: float = 0.1, level_down_threshold: float = 150.0, level_up_threshold: float = 750.0, max_scale: float = 1.0, min_scale: float = 0.0, num_episode_average: int = 1000, reward_term_names: list[str] = [], reward_term_tags: list[str] = [], tracker_name: str \| None = None	Curriculum that gradually scales reward weights for a set of reward terms.
stage_reward_weight_curriculum	class StageRewardWeightCurriculum	reward_term_name: str = '', weight_stages: list[tuple[int, float]] = []	Curriculum that sets a reward term's weight at given training steps.

eden.managers.terms.curricula.success_guided

Term	Form	Parameters	Summary
success_guided_sampling	class SuccessGuidedSampling	concentration: float = 5.0, floor: float = 1e-08, history: int = 100, success_term: str \| None = None, target: float = 0.66, temperature: float = 2.0	Draw each resetting env's reset configuration with probability peaked at a target success rate.

eden.managers.terms.curricula.velocity

Term	Form	Parameters	Summary
velocity_command_curriculum	class VelocityCommandCurriculum	command_name: str = 'velocity_command', velocity_stages: list[dict] = []	Curriculum that progressively expands velocity command ranges.

metric terms

20 registered.

• Function terms: MetricTerm.configure(func=metrics.<term>, params={...}) — from eden import metrics, from eden.terms import MetricTerm
• Class terms: <ClassName>.configure(<field>=...) — import it from the module its section names
• Every metric term also accepts: contributes_to_success: bool = True, direction: MetricDirection = MetricDirection.HIB, is_cumulative: bool = False, metric_mode: MetricMode = MetricMode.INTERVAL, success_threshold: float = 1.0

eden.managers.terms.metrics.articulation

Term	Form	Parameters	Summary
joint_angle_at_target	function	entity_name: str, joint_name: str, target_angle: float, threshold: float = 0.05	Check if a single DOF is within threshold of target_angle.

eden.managers.terms.metrics.boolean

Term	Form	Parameters	Summary
metric_term_and	class MetricTermAND	terms: list[MetricTermOptions] = []	Metric term that returns the element-wise minimum of its child terms.
metric_term_not	class MetricTermNOT	term: MetricTermOptions = None	Metric term that inverts a single child term (1.0 - value).
metric_term_or	class MetricTermOR	terms: list[MetricTermOptions] = []	Metric term that returns the element-wise maximum of its child terms.

eden.managers.terms.metrics.containment

Term	Form	Parameters	Summary
object_above_height	function	entity_name: str, height: float, relative: bool = False	Check if an entity's root z-position is above height.
object_bottom_above_height	function	entity_name: str, height: float	Lowest point of the entity above height.
object_in_region	function	entity_name: str, region_min: list[float], region_max: list[float], relative: bool = False	Check if an entity's root position is inside an axis-aligned box.

eden.managers.terms.metrics.error

Term	Form	Parameters	Summary
error_vel_xy	function	command_name: str = 'velocity'	—
error_vel_yaw	function	command_name: str = 'velocity'	—

eden.managers.terms.metrics.grasp

Term	Form	Parameters	Summary
is_grasping	class IsGrasping	force_threshold: float = 1.0, left_gripper_link_name: str = 'left_gripper_pad', right_gripper_link_name: str = 'right_gripper_pad', robot_name: str = 'robot'	Detect a parallel-jaw grasp via contact force alignment.
object_lifted	function	entity_name: str, min_height: float	Check if an entity has been lifted above min_height.

eden.managers.terms.metrics.proximity

Term	Form	Parameters	Summary
ee_near_entity	class EeNearEntity	ee_link_names: list[str] = [], entity_name: str = 'obj', robot_name: str = 'robot', threshold: float = 0.05	Check if the mean end-effector position is within threshold of a target entity.
entity_distance	function	entity_a_name: str, entity_b_name: str	Distance between two entity root positions.
entity_distance_xy	function	entity_a_name: str, entity_b_name: str	Horizontal (XY) distance between two entity root positions.
entity_near_entity_xy	function	entity_a_name: str, entity_b_name: str, threshold: float = 0.05	Check if two entities are within threshold horizontal (XY) distance.
entity_near_target	function	entity_name: str, target_pos: Vec3FType \| torch.Tensor, threshold: float = 0.05	Check if an entity is within threshold of a fixed world position.
object_a_is_on_b	function	entity_a_name: str, entity_b_name: str, xy_threshold: float = 0.03, height_threshold: float = 0.04, height_diff: float = 0.0	Check if object_a is on object_b by comparing height and xy distance of object origins.
object_a_within_b_xy	function	entity_a_name: str, entity_b_name: str, min_fraction: float = 0.8	min_fraction of entity_a's footprint lies over entity_b's.

eden.managers.terms.metrics.sequence

Term	Form	Parameters	Summary
rubric_metric_term	class RubricMetricTerm	criteria: list[CriterionOptions] = []	Fraction of rubric criteria met, each after its dependencies.
sequential_metric_term	class SequentialMetricTerm	allow_skip: bool = False, phases: list[PhaseOptions] = []	Metric that tracks ordered sub-goals (phases).

recorder terms

7 registered.

• Class terms: <ClassName>.configure(<field>=...) — import it from the module its section names

eden.managers.terms.recorders.actions

Term	Form	Parameters	Summary
action_recorder	class ActionRecorder	—	—

eden.managers.terms.recorders.observations

Term	Form	Parameters	Summary
observation_recorder	class ObservationRecorder	groups: list[str] \| None = None	Record observation groups at the start of each step, as obs/<group>.

eden.managers.terms.recorders.rl

Term	Form	Parameters	Summary
rl_step_recorder	class RLStepRecorder	include_done: bool = True, include_reward: bool = True, include_terminated: bool = True, include_timeout: bool = True, include_timestamp: bool = True	Record per-step RL buffers (reward, done, terminated, timeout, timestamp).

eden.managers.terms.recorders.sensors

Term	Form	Parameters	Summary
sensor_recorder	class SensorRecorder	record_post_reset_snapshot: bool = True, use_ground_truth: bool = False	Record sensor outputs during simulation.

eden.managers.terms.recorders.states

Term	Form	Parameters	Summary
initial_state_recorder	class InitialStateRecorder	include_dofs_vel: bool = True, include_pos_quat: bool = True, include_qpos: bool = True	Record per-episode initial entity states right after reset.
state_recorder	class StateRecorder	entity_name: str \| list[str] = 'robot', include_dofs_vel: bool = True, include_pos_quat: bool = True, include_qpos: bool = True	Record the state of the environment at the beginning of each step.

eden.managers.terms.recorders.sysid

Term	Form	Parameters	Summary
sys_id_recorder	class SysIDRecorder	entity_name: str = 'robot', include_base: bool = True, include_torque: bool = True	Record the proprioceptive trace needed to rebuild a sysid Trajectory.

motion terms

1 registered.

• Class terms: <ClassName>.configure(<field>=...) — import it from the module its section names
• Every motion term also accepts: adaptive_alpha: float = 0.001, adaptive_failure_rate_max_over_mean: float = 10.0, adaptive_kernel_size: int = 1, adaptive_lambda: float = 0.8, adaptive_motion_weights: bool = False, adaptive_sampling: bool = False, adaptive_uniform_ratio: float = 0.1, compute_per_motion_metrics: bool = False, dataset_options: DatasetOptions \| None = None, entities_map: dict[str, str] = {}, env_motion_groups: MotionGroupOptions \| None = None, env_starting_stub_sampling_strategy: Literal['independent', 'concat_motion_bins'] = 'independent', freeze_at_timestep_zero_prob: float = 0.0, log_motion_id_to_filename: bool = False, log_per_motion_metrics: bool = False, metric_anchor_link_name: str \| None = None, motion_bin_length_s: float \| None = None, motion_start_from_middle_range: tuple[float, float] \| list[float] = (0.0, 0.0), motion_weight_failure_amplify: float = 2.0, motion_weight_failure_rate_max_over_mean: float = 200.0, motion_weight_floor: float = 0.001, motion_weight_success_discount: float = 0.999, motion_weight_uniform_mix: float = 0.0, motion_weight_update_every_n_iters: int = 1, motion_weight_update_mode: Literal['compound', 'snapshot'] = 'snapshot', motion_weights: str \| list[float] \| None = None, start_at_timestep_zero_prob: float = 0.0, target_dofs_name: dict[str, str \| list[str]] = {}, target_links_name: dict[str, str \| list[str]] = {}

eden.managers.motion_manager

Term	Form	Parameters	Summary
motion_term	class MotionTerm	—	—

noise_model terms

3 registered.

• Class terms: <ClassName>.configure(<field>=...) — import it from the module its section names
• Every noise_model term also accepts: operation: NoiseOperation = NoiseOperation.ADD

eden.managers.modifiers.observations.noise

Term	Form	Parameters	Summary
constant_noise	class ConstantNoise	bias: torch.Tensor \| float = 0.0	—
gaussian_noise	class GaussianNoise	mean: torch.Tensor \| float = 0.0, std: torch.Tensor \| float = 1.0	—
uniform_noise	class UniformNoise	n_max: torch.Tensor \| float = 1.0, n_min: torch.Tensor \| float = -1.0	—

observation_modifier terms

2 registered.

• Class terms: <ClassName>.configure(<field>=...) — import it from the module its section names

eden.managers.modifiers.observations.temporal

Term	Form	Parameters	Summary
exponential_moving_average	class ExponentialMovingAverage	alpha: float \| None = None, backfill: bool = True	Exponential moving average, updated once per control step.
observation_delay	class ObservationDelay	backfill: bool = True, control_steps: int \| None = None	Delivers the frame the term produced control_steps control steps earlier.

action_modifier terms

10 registered.

• Class terms: <ClassName>.configure(<field>=...) — import it from the module its section names

eden.managers.modifiers.actions.actuators

Term	Form	Parameters	Summary
action_delay	class ActionDelay	delay_range_s: tuple[float, float] \| None = None, delay_range_steps: tuple[int, int] \| None = None	Delays the command the actuator acts on by a random number of physics steps, drawn per episode.
compose	class Compose	modifiers: tuple[ActionModifierOptions, ...] \| list[ActionModifierOptions] = ()	Chains multiple action modifiers in sequence.
constant_torque_kick	class ConstantTorqueKick	activation_epsilon: LaxFArrayType = 0.0, direction_source: str = 'pos_err', torque_kick: LaxFArrayType = 0.0	Adds a flat directional torque on top of computed control torque.
deadband	class Deadband	deadband_epsilon: LaxFArrayType = 0.0	Zeroes PD control torque inside a position-error deadband.
effort_clip	class EffortClip	braking_torque_limit: LaxFArrayType \| None = None, driving_torque_limit: LaxFArrayType \| None = None, full_torque_speed: LaxFArrayType \| None = None, no_load_speed: LaxFArrayType \| None = None	Clips torques to motor torque-speed (T-N) curve limits.
envelope_clip	class EnvelopeClip	braking_torque_limit: LaxFArrayType \| None = None, driving_torque_limit: LaxFArrayType \| None = None, full_torque_speed: LaxFArrayType \| None = None, no_load_speed: LaxFArrayType \| None = None	Clips target positions so PD torques stay within the motor torque-speed envelope.
friction_model	class FrictionModel	dynamic_friction: LaxFArrayType \| None = None, friction_activation_vel: LaxFArrayType \| None = None, friction_offset: LaxFArrayType = 0.0, static_friction: LaxFArrayType \| None = None	Applies tanh-based static + viscous dynamic friction to torques.
gear_backlash	class GearBacklash	backlash: LaxFArrayType = 0.0, initial_side: LaxFArrayType = 0.0, reversal_threshold: LaxFArrayType = 0.0, takeup_rate: LaxFArrayType = 1.0	Adds direction-dependent gear slop to position targets before PD control.
motor_strength	class MotorStrength	motor_strength: LaxFArrayType = 1.0	Scales control torques by a motor strength multiplier.
torque_offset	class TorqueOffset	torque_offset: LaxFArrayType = 0.0	Adds a constant offset to the control torques.