deepMind Lab 可以调用的python API 文档 ，从deepMindlab的代码上直接复制粘贴的，方便今后查看。

Python environment API

The Python module deepmind_lab defines the Lab class. For example usage, there is python/tests/dmlab_module_test.py,python/random_agent.py, and python/random_agent_simple.py.

class deepmind_lab.Lab(level, observations, config={}, renderer='software')

Creates an environment object, loading the game script file level. The environment's observations() method will return the observations specified by name in the list observations.

The config dict specifies additional settings as key-value string pairs. The following options are recognized:

Option	Description	Default
width	horizontal resolution of the observation frames	'320'
height	vertical resolution of the observation frames	'240'
fps	frames per second	'60'
appendCommand	Commands for the internal Quake console*	''

* See also Lua map API.

Unrecognized options are passed down to the level's init function. In Lua, this is kwargs.opts in api:init.

For example, you can run a modified version of the lt_chasm level with these calls,

import deepmind_lab

observations = ['RGBD']
env = deepmind_lab.Lab('lt_chasm', observations,
                       config={
  'width':