pycram.external_interfaces.pycramgym#

Classes#

PyCRAMGym

First small implementation of the PyCRAMGYM, this class can either be used to store necessary information for an experiment or let

Module Contents#

class pycram.external_interfaces.pycramgym.PyCRAMGym(inWorld: str, inRobot: str, inPlan: pycram.language.LanguageMixin)#

Bases: gymnasium.Env

First small implementation of the PyCRAMGYM, this class can either be used to store necessary information for an experiment or let a robot explore an environment, with a set of actions. The exploration will be done via reinforcement learning.

world: str#

The world should be an real World Object, but sadly this is not possible, because if we close and open multiple bulletworlds in quick succession, then there is a problem with objects not being loaded properly. Thus we decided to let the user create the world they want to use and just load the environment urdf of the file they give to this class

robot: str#

The robot should be a real Robot Object, but in order to stay in line with the world it is just the name of the urdf that will be loaded as a robot.

plan: pycram.language.LanguageMixin#

Plan langauge object that can get executed.

step(action: gymnasium.core.ActType)#

Ticks the world, in our case this means we pick one possible action and perform it.

Variables:

action – The action to take.

reset()#

Returns the environment to the normal state.

close()#

Resets the world to a completely empty state. (Should close world at some point)