bolero.representation.DMPSequence¶

class bolero.representation.DMPSequence(n_dmps=1, execution_times=None, dt=0.01, n_features=None, subgoals=None, learn_goal_velocities=False, initial_weights=None)[source]¶

Sequence of DMPs.

Each DMP is initialized at the last phase of its predecessor to ensure smooth transitions.

Parameters:

n_dmps : int, optional (default: 1) Number of DMPs in this sequence execution_times : array-like, shape (n_dmps,), optional (default: ones) Execution times of the DMPs n_features : array-like, shape (n_dmps,), optional (default: 50s) Number of RBF features for each dimension of each DMP. dt : float, optional (default: 0.01) Time between successive steps in seconds. subgoals : array-like, shape (n_dmps + 1, n_task_dims) Subgoals of the DMPs including the final goal and the initial state. learn_goal_velocities : bool, optional (default: False) Defines whether the parameter vector will include the velocities at the goals of the DMPs. initial_weights : list List of initial weight vectors for the DMPs. Each entry initial_weights[i] has the shape n_features[i] * n_task_dims

__init__(n_dmps=1, execution_times=None, dt=0.01, n_features=None, subgoals=None, learn_goal_velocities=False, initial_weights=None)[source]¶

can_step()[source]¶

Returns true if step() can be called again, false otherwise.

get_args()¶

Get parameters for this estimator.

Returns:	params : mapping of string to any Parameter names mapped to their values.

get_n_params()[source]¶

Get number of parameters.

get_outputs(outputs)[source]¶

Get desired next system state.

Parameters:	outputs : array-like, shape (n_task_dims * 3) Position, velocitie and rotation of each dimension

get_params()[source]¶

Utility function: get currently optimizable parameters.

get_subgoal(idx)[source]¶

Get subgoal.

Parameters:	idx: int index of the subgoal: 0 is the start of the DMP sequence, -1 would be the goal of the sequence
Returns:	subgoal: array-like, shape (n_task_dims,) subgoal at index ‘idx’

get_subgoal_velocity(idx)[source]¶

Get subgoal.

Parameters:	idx: int index of the subgoal: 0 is the start of the DMP sequence, -1 would be the goal of the sequence
Returns:	subgoal: array-like, shape (n_task_dims,) velocity at subgoal with index ‘idx’

init(n_inputs, n_outputs)[source]¶

Initialize the behavior.

Parameters:	n_inputs : int number of inputs n_outputs : int number of outputs

reset()[source]¶

Reset DMP.

set_inputs(inputs)[source]¶

Set current system state.

Parameters:	inputs : array-like, shape (n_task_dims * 3) Position, velocitie and rotation of each dimension

set_meta_parameters(keys, meta_parameters)[source]¶

Set DMP meta parameters.

Required meta-parameters

x0 : array

initial position

g : array

goal

Parameters:	keys : list of string names of meta-parameters meta_parameters : list of float values of meta-parameters

set_params(params)[source]¶

Utility function: set currently optimizable parameters.

set_subgoal(idx, subgoal)[source]¶

Set subgoal manually.

Parameters:	idx: int index of the subgoal: 0 is the start of the DMP sequence, -1 would be the goal of the sequence subgoal: array-like, shape (n_task_dims,) subgoal at index ‘idx’

set_subgoal_velocity(idx, subgoal_vel)[source]¶

Set subgoal velocity manually.

Parameters:	idx: int index of the subgoal: 0 is the start of the DMP sequence, -1 would be the goal of the sequence subgoal_vel: array-like, shape (n_task_dims,) velocity at subgoal with index ‘idx’

step()[source]¶

Compute next step.

trajectory()[source]¶

Generate trajectory represented by the sequence of DMPs in open loop.

The function can be used for debugging purposes.

Returns:	X : array, shape (n_steps, n_task_dims) Positions Xd : array, shape (n_steps, n_task_dims) Velocities Xdd : array, shape (n_steps, n_task_dims) Accelerations