Robust Maze Tasks

Robust Maze Tasks

TasksRobust type	Robust State	Robust Action	Robust Reward
AntMaze_UMaze-v4	✅	✅	✅
PointMaze_UMaze-v3	✅	✅	✅

A Simple Example

import robust_gymnasium as gym
import numpy as np
import json
import os
import time
from datetime import datetime

# Get current date and time for folder naming
currentDateAndTime = datetime.now()
start_run_date_and_time = time.strftime("%Y-%m-%d-%H-%M-%S", time.localtime())

# Import configuration settings
from robust_gymnasium.configs.robust_setting import get_config
args = get_config().parse_args()

# Set environment and noise factor
args.env_name = "AntMaze_UMaze-v4"
args.noise_factor = "state"

# Define folder path for storing data
folder = os.getcwd()[:0] + 'data/' + str(args.env_name) + '/' + str(args.noise_type) + '/' + str(
    start_run_date_and_time) + '/'
print("folder---:", folder)

# Create folder if it doesn't exist
if not os.path.exists(folder):
    os.makedirs(folder)

# Save configuration settings to a JSON file
json_path = folder + '/config.json'
argsDict = args.__dict__
with open(json_path, 'w') as f:
    f.writelines('------------------ start ------------------' + '\n')
    for eachArg, value in argsDict.items():
        f.writelines(eachArg + ' : ' + str(value) + '\n')
    f.writelines('------------------- end -------------------')

# Function to test AntMaze environment reset behavior
def test_reset():
    """Check that AntMaze does not reset into a success state."""
    env = gym.make("AntMaze_UMaze-v4", continuing_task=True, render_mode="human")

    for _ in range(10000):
        obs, info = env.reset()
        assert not info["success"]
        dist = np.linalg.norm(obs["achieved_goal"] - obs["desired_goal"])
        env.render()  # Render environment
        assert dist > 0.45, f"dist={dist} < 0.45"

if __name__ == '__main__':
    test_reset()