refine_plan.algorithms.refine
=============================

.. py:module:: refine_plan.algorithms.refine

.. autoapi-nested-parse::

   Functions which bring the stages of BT refinement together.

   There are two key stages to refinement:
   1. State space extraction and option learning.
   2. Semi-MDP construction and solution, and policy->BT conversion.

   These two stages are represented with different functions.

   Author: Charlie Street
   Owner: Charlie Street



Functions
---------

.. autoapisummary::

   refine_plan.algorithms.refine.synthesise_bt_from_options


Module Contents
---------------

.. py:function:: synthesise_bt_from_options(sf_list, option_list, labels, initial_state=None, prism_prop='Rmax=?[F "goal"]', default_action='None', out_file=None, bt_converter='normal')

   Synthesise a BT using options learned from the initial BT.

   :param sf_list: A list of StateFactor objects for the semi-MDP
   :param option_list: A list of Option objects which capture robot capabilities
   :param labels: A list of Label objects within the semi-MDP
   :param initial_state: The semi-MDP initial state (if there is one)
   :param prism_prop: The PRISM property to solve for (passed to Storm)
   :param default_action: The action which replaces any None actions in a policy
   :param out_file: Optional. The path for the final BT XML file
   :param bt_converter: Can be normal (PolicyBTConverter) or simple (SimplePolicyBTConverter)

   :returns: The final refined BT