Simulation environment of task with autonomous mobile robot using Navigation Stack.

In this environment, the agent do action of instructing relative navigation goal pose and observe a subjective occupancy map.

TreasureChestRoom:

Agent aim to open chests in unknown rooms with keys and discover as much treasure as possible.

The rooms in which agent spawned are randomly generated such as the following structure.

Yellow cube is key, and cyan cube is treasure chest. Each object will be generated based on different set of placing rules.

For more information, please click here: navstack-gym/__init__.py

pip install navstack-gym

Example of Learning Code with DrQ Agent:
wwwshwww/navstack-gym-expr

Random Walk Script:

import gym
import navstack_gym

env = gym.make('VisibleTreasureHunt-v0')
obs = env.reset(is_generate_pose=True, is_generate_room=True, obstacle_count=10)

imgs = []
imgs.append(env.render('rgb_array'))

for i in range(10):
    action = env.action_space.sample()
    obs, reward, done, info = env.step(action)
    imgs.append(env.render('rgb_array'))