Welcome to Anti-Poaching Environment (APE)’s documentation!

Fig. 1 Instance of APE with 2 Rangers (blue) vs 4 Poachers (red)

APE is a PettingZoo environment that implements the Anti-Poaching Game. This is a multi-agent, zero-sum, cooperative-competitive game played between a team of Rangers and some independent Poachers on a grid.

We first quickly introduce Anti-Poaching Environment (APE) and how to use it with the given examples. A more detailed explanation of the examples, notably the integration with RLlib is given in Examples. implementation lists the entire API for APE and the RLlib integration.

• Anti-Poaching Environment (APE)
  • Installation
  • Using APE
  • Examples and the RLlib Interface
    • Manual policies
    • Rllib examples
• Examples
  • Manual Examples
  • RLlib Examples
    • main.py
    • Visualising Saved Policies
    • Hyper-parameter Tuning
• The APE API
  • anti_poaching.env package
    • Subpackages
      • anti_poaching.env.utils package
        • Submodules
        • anti_poaching.env.utils.game_utils module
          • BaseGridState
            • BaseGridState.NULL_POS
            • BaseGridState.RENDER_MODES
            • BaseGridState.REWARD_MAP
            • BaseGridState.add_trap()
            • BaseGridState.get_neighbours()
            • BaseGridState.permitted_movements()
            • BaseGridState.prob_animal_appear()
            • BaseGridState.prob_detect_cell()
            • BaseGridState.prob_detect_trap()
            • BaseGridState.remove_poacher()
            • BaseGridState.remove_trap()
            • BaseGridState.render()
            • BaseGridState.reset()
            • BaseGridState.update_position()
            • BaseGridState.valid_pos()
          • GridStateConstProb
            • GridStateConstProb.permitted_movements()
            • GridStateConstProb.prob_animal_appear()
            • GridStateConstProb.prob_detect_cell()
            • GridStateConstProb.prob_detect_trap()
            • GridStateConstProb.update_position()
            • GridStateConstProb.valid_pos()
          • GridStateVaryingPCA
            • GridStateVaryingPCA.permitted_movements()
            • GridStateVaryingPCA.prob_animal_appear()
            • GridStateVaryingPCA.prob_detect_cell()
            • GridStateVaryingPCA.prob_detect_trap()
            • GridStateVaryingPCA.update_position()
            • GridStateVaryingPCA.valid_pos()
          • Trap
            • Trap.name
            • Trap.value
        • anti_poaching.env.utils.typing module
        • anti_poaching.env.utils.wrappers module
          • FullStateObservationWrapper
            • FullStateObservationWrapper.observation()
          • QMIXCompatibilityLayer
            • QMIXCompatibilityLayer.action_space
            • QMIXCompatibilityLayer.get_sub_environments
            • QMIXCompatibilityLayer.observation()
            • QMIXCompatibilityLayer.observation_space
            • QMIXCompatibilityLayer.reset()
            • QMIXCompatibilityLayer.step()
        • Module contents
    • Submodules
    • anti_poaching.env.anti_poaching module
      • parallel_env()
      • raw_env
        • raw_env.action_space()
        • raw_env.action_spaces
        • raw_env.agents
        • raw_env.metadata
        • raw_env.observation_space()
        • raw_env.observation_spaces
        • raw_env.possible_agents
        • raw_env.render()
        • raw_env.reset()
        • raw_env.state()
        • raw_env.step()
        • raw_env.tap()
    • Module contents
  • anti_poaching.env.utils package
    • Submodules
    • anti_poaching.env.utils.game_utils module
      • BaseGridState
        • BaseGridState.NULL_POS
        • BaseGridState.RENDER_MODES
        • BaseGridState.REWARD_MAP
        • BaseGridState.add_trap()
        • BaseGridState.get_neighbours()
        • BaseGridState.permitted_movements()
        • BaseGridState.prob_animal_appear()
        • BaseGridState.prob_detect_cell()
        • BaseGridState.prob_detect_trap()
        • BaseGridState.remove_poacher()
        • BaseGridState.remove_trap()
        • BaseGridState.render()
        • BaseGridState.reset()
        • BaseGridState.update_position()
        • BaseGridState.valid_pos()
      • GridStateConstProb
        • GridStateConstProb.permitted_movements()
        • GridStateConstProb.prob_animal_appear()
        • GridStateConstProb.prob_detect_cell()
        • GridStateConstProb.prob_detect_trap()
        • GridStateConstProb.update_position()
        • GridStateConstProb.valid_pos()
      • GridStateVaryingPCA
        • GridStateVaryingPCA.permitted_movements()
        • GridStateVaryingPCA.prob_animal_appear()
        • GridStateVaryingPCA.prob_detect_cell()
        • GridStateVaryingPCA.prob_detect_trap()
        • GridStateVaryingPCA.update_position()
        • GridStateVaryingPCA.valid_pos()
      • Trap
        • Trap.name
        • Trap.value
    • anti_poaching.env.utils.typing module
    • anti_poaching.env.utils.wrappers module
      • FullStateObservationWrapper
        • FullStateObservationWrapper.observation()
      • QMIXCompatibilityLayer
        • QMIXCompatibilityLayer.action_space
        • QMIXCompatibilityLayer.get_sub_environments
        • QMIXCompatibilityLayer.observation()
        • QMIXCompatibilityLayer.observation_space
        • QMIXCompatibilityLayer.reset()
        • QMIXCompatibilityLayer.step()
    • Module contents
• Integration with RLlib
  • rllib package
    • Submodules
    • rllib.callbacks module
      • CollectMetricsAndExploit
      • EpisodeMetricsCallbacks
        • EpisodeMetricsCallbacks.on_episode_start()
        • EpisodeMetricsCallbacks.on_episode_step()
      • EvaluationSaveCallback
        • EvaluationSaveCallback.on_evaluate_start()
      • ExploitabilityCallback
        • ExploitabilityCallback.on_evaluate_end()
      • RestoreNonTrainingAgents
        • RestoreNonTrainingAgents.on_algorithm_init()
      • SumRangerRewardMetric
        • SumRangerRewardMetric.on_train_result()
    • rllib.configs module
      • duplicator()
      • generate_training_config()
      • generic_conf()
      • pg_config()
      • ppo_config()
      • qmix_config()
      • read_hyperparams_for_algo()
    • rllib.example_utils module
      • define_game_from_args()
      • get_args()
      • trial_path_generator()
    • rllib.from_checkpoint module
      • get_actions()
    • rllib.heuristic_policies module
      • ActionMaskedGreedyPoacherPolicy
        • ActionMaskedGreedyPoacherPolicy.compute_actions()
        • ActionMaskedGreedyPoacherPolicy.get_initial_state()
        • ActionMaskedGreedyPoacherPolicy.get_weights()
        • ActionMaskedGreedyPoacherPolicy.is_recurrent()
        • ActionMaskedGreedyPoacherPolicy.learn_on_batch()
        • ActionMaskedGreedyPoacherPolicy.set_weights()
      • ActionMaskedRandomPolicy
        • ActionMaskedRandomPolicy.compute_actions()
        • ActionMaskedRandomPolicy.get_initial_state()
        • ActionMaskedRandomPolicy.get_weights()
        • ActionMaskedRandomPolicy.is_recurrent()
        • ActionMaskedRandomPolicy.learn_on_batch()
        • ActionMaskedRandomPolicy.set_weights()
      • ActionMaskedStaticPoacherPolicy
        • ActionMaskedStaticPoacherPolicy.compute_actions()
        • ActionMaskedStaticPoacherPolicy.compute_single_action()
        • ActionMaskedStaticPoacherPolicy.get_initial_state()
        • ActionMaskedStaticPoacherPolicy.get_weights()
        • ActionMaskedStaticPoacherPolicy.is_recurrent()
        • ActionMaskedStaticPoacherPolicy.learn_on_batch()
        • ActionMaskedStaticPoacherPolicy.set_weights()
      • QMIXActionMaskedRandomPolicy
        • QMIXActionMaskedRandomPolicy.compute_actions_from_input_dict()
      • QMIXBaseHeuristicPolicy
        • QMIXBaseHeuristicPolicy.compute_actions()
        • QMIXBaseHeuristicPolicy.compute_actions_from_input_dict()
        • QMIXBaseHeuristicPolicy.learn_on_batch()
      • QMIXGreedyStaticPolicy
        • QMIXGreedyStaticPolicy.compute_actions_from_input_dict()
    • rllib.main module
      • get_policies_from_arg()
    • rllib.recurrent module
      • LSTMActionMaskingModel
        • LSTMActionMaskingModel.forward()
        • LSTMActionMaskingModel.forward_rnn()
        • LSTMActionMaskingModel.get_initial_state()
        • LSTMActionMaskingModel.value_function()
      • QmixLSTMModel
        • QmixLSTMModel.forward()
        • QmixLSTMModel.get_initial_state()
        • QmixLSTMModel.value_function()
    • rllib.tune_antipoaching module
    • Module contents
• CHANGELOG
  • Summary
  • v0.3
  • v0.2.2
  • v0.2.1
  • v0.2.0
    • Bugfixes

Indices and tables

• Index

• Module Index

• Search Page