example_envs.tag_gridworld package
Submodules
example_envs.tag_gridworld.tag_gridworld module
- class example_envs.tag_gridworld.tag_gridworld.CUDATagGridWorld(num_taggers=10, grid_length=10, episode_length=100, starting_location_x=None, starting_location_y=None, seed=None, wall_hit_penalty=0.1, tag_reward_for_tagger=10.0, tag_penalty_for_runner=2.0, step_cost_for_tagger=0.01, use_full_observation=True)
Bases:
example_envs.tag_gridworld.tag_gridworld.TagGridWorld
,warp_drive.utils.gpu_environment_context.CUDAEnvironmentContext
CUDA version of the TagGridWorld environment. Note: this class subclasses the Python environment class TagGridWorld, and also the CUDAEnvironmentContext
- get_data_dictionary()
- get_tensor_dictionary()
- step(actions=None)
- class example_envs.tag_gridworld.tag_gridworld.TagGridWorld(num_taggers=10, grid_length=10, episode_length=100, starting_location_x=None, starting_location_y=None, seed=None, wall_hit_penalty=0.1, tag_reward_for_tagger=10.0, tag_penalty_for_runner=2.0, step_cost_for_tagger=0.01, use_full_observation=True)
Bases:
object
The game of tag on a 2D square grid plane. This is a simplified version of the continuous tag. There are a number of taggers trying to tag 1 runner. The taggers want to catch the runner. Once the runner is tagged, the game is over.
- generate_observation()
- name = 'TagGridWorld'
- reset()
- seed(seed=None)
- set_global_state(key=None, value=None, t=None, dtype=None)
- step(actions=None)
- update_state(actions_x, actions_y)