当前位置: 首页>>代码示例>>Python>>正文


Python Env.step方法代码示例

本文整理汇总了Python中env.Env.step方法的典型用法代码示例。如果您正苦于以下问题:Python Env.step方法的具体用法?Python Env.step怎么用?Python Env.step使用的例子?那么恭喜您, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在env.Env的用法示例。


在下文中一共展示了Env.step方法的1个代码示例,这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞,您的评价将有助于系统推荐出更棒的Python代码示例。

示例1: Field

# 需要导入模块: from env import Env [as 别名]
# 或者: from env.Env import step [as 别名]
class Field(ScatterPlane):
    '''
    This is the Field which will contain cells.
    '''
    agent_widget = ObjectProperty(None)
    total_reward = NumericProperty(0)

    def __init__(self, cell_size=25, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self.cell_size = cell_size

        # At the __init__ height and width, and consecutively center may be not
        # established, yet due to layout logic.
        Clock.schedule_once(self._init_after)

        Clock.schedule_interval(self.update, 0.1)

    def _init_after(self, dt):
        ''' Perform initializations after the layout is finalized. '''
        self.env = Env()
        # TODO: Move params to config file
        with open('sarsa.pickle', 'rb') as fd:
            self.sarsa = pickle.load(fd)
        self.grid = Grid(self.canvas, 'line_loop', Color(), self.cell_size,
                         self.to_local(*self.center))
        self.state = self.env.reset(self.grid)
        self._place_agent(self.state.cell)

    def _place_agent(self, cell):
        self.agent_widget.center = self.grid.pixcenter(cell.q, cell.r)
        # FIXME
        for _ in self.grid.neighbors(cell.q, cell.r):
            pass

    def on_touch_down(self, touch):
        super().on_touch_down(touch)

        x, y = self.to_local(touch.x, touch.y)
        q, r = self.grid.pixel_to_hex(x, y)

        if (q, r) in self.grid:
            print("Touched ({}, {}) in {}.".format(q, r, (x, y)))
            print("env tvisited", self.env.tvisited[q, r])
            print("state food", self.state.food)
        else:
            self.grid.init(q, r)
            for _ in self.grid.neighbors(q, r):
                pass

        return True

    # TODO: Shouldn't this feel better in SwarmApp?
    def update(self, dt):
        action = self.sarsa.policy(self.state, explore=False)
        next_state, reward, done = self.env.step(action)
        self.sarsa.adapt_policy(self.state, action, next_state, reward)

        self.state = next_state
        self.total_reward += int(reward)
        self._place_agent(self.state.cell)
开发者ID:Uran198,项目名称:kivy_swarm,代码行数:62,代码来源:main.py


注:本文中的env.Env.step方法示例由纯净天空整理自Github/MSDocs等开源代码及文档管理平台,相关代码片段筛选自各路编程大神贡献的开源项目,源码版权归原作者所有,传播和使用请参考对应项目的License;未经允许,请勿转载。