Returns the positions of all agents in a group. Args: handle (ctypes.c_int32): Group handle. Returns: pos (n

Set parameters of global configuration. :param dict args: Contains the following configuration attributes: * **map_width

Add a reward rule. Args: on (Event): An objecting representing a bool expression of the trigger event. receiver (List

Add a group to the configuration. Args: agent_type (str): Name of agent type contained in this group. Returns:

Adds agents to environment. Args: handle (ctypes.c_int32): The handle of the group to which to add the agents. method

Register an agent type. :param str name: Name of the type (should be unique). :param dict attr: Contains the following confi

Returns all group handles in the environment. Returns: handles (List[ctypes.c_int32]): All group handles in the environment.

Returns the number of agents in a group. Args: handle (ctypes.c_int32): Group handle. Returns: num (int): Nu

numpy to ctypes array

Adds walls to the environment. Args: method (str): Can be 'random' or 'custom'. If method is 'random', then kwargs["n"] is an int

numpy to ctypes array

Returns the ids of all agents in the group. Args: handle (ctypes.c_int32): Group handle. Returns: ids (np.nd

private method, for interactive application

get buffer to receive observation from c++ engine

private method, for interactive application

Returns the alive status of all agents in a group. Args: handle (ctypes.c_int32): Group handle. Returns: ali

Returns the feature space for a group. Args: handle (ctypes.c_int32): Group handle. Returns: feature_space (

Returns the observation for each agent in a group. Args: handle (ctypes.c_int32): Group handle. Returns: obs

Returns the view space for a group. Args: handle (ctypes.c_int32): Group handle. Returns: view_space (Tuple[

* GridWorld special */ agent

private method, for interactive application

init observation buffer

Load library in local.

serialize event expression and sent them to game engine

Clears dead agents in the engine. Must be called after step().

* some interface for python bind */

do attack for agent, return kill_reward and dead_group

do move for agent, dx = delta[0], dy = delta[1]

to turn, wise = 1 for clockwise, wise = -1 for counter-clockwise

download url to filename

download model from url

* General Environment */

render