bridgewalk 0.1.0

Visual reinforcement learning benchmark for controllability.

BridgeWalk

BridgeWalk is a partially-observed reinforcement learning environment with dynamics of varying stochasticity. The player needs to walk along a bridge to reach a goal location. When the player walks off the bridge into the water, the current will move it randomly until it gets washed back on the shore. A good agent in this environment avoids this stochastic trap. The implementation of BridgeWalk is based on the Crafter environment.

Play Yourself

You can play the game yourself with an interactive window and keyboard input. The mapping from keys to actions, health level, and inventory state are printed to the terminal.

# Install with GUI
pip3 install 'bridgewalk[gui]'

# Start the game
bridgewalk

# Alternative way to start the game
python3 -m bridgewalk.run_gui

The following optional command line flags are available:

Flag	Default	Description
--window <width> <height>	800 800	Window size in pixels, used as width and height.
--fps <integer>	5	How many times to update the environment per second.
--record <filename>.mp4	None	Record a video of the trajectory.
--view <width> <height>	7 7	The layout size in cells; determines view distance.
--length <integer>	None	Time limit for the episode.
--seed <integer>	None	Determines world generation and creatures.

Training Agents

Installation: pip3 install -U bridgewalk

The environment follows the OpenAI Gym interface:

import bridgewalk

env = bridgewalk.Env(seed=0)
obs = env.reset()
assert obs.shape == (64, 64, 3)

done = False
while not done:
  action = env.action_space.sample()
  obs, reward, done, info = env.step(action)

Environment Details

Reward

A reward of +1 is given the first time in each episode when the agent reaches the island at the end of the bridge.

Termination

Episodes terminate after 250 steps.

Observation Space

Each observation is an RGB image that shows a local view of the world around the player, as well as the inventory state of the agent.

Action Space

The action space is categorical. Each action is an integer index representing one of the possible actions:

Integer	Name	Description
0	noop	Do nothing.
1	move_left	Walk left.
2	move_right	Walk right.
3	move_up	Walk up.
4	move_down	Walk down.

Questions

Please open an issue on Github.