Warning (2022.05.06):

@@ I will stop further development on this repo as Mujoco is now free. @@
@@ However I will keep maintaining the repo and respond to issues. @@
@@ Feel free to keep using the package and contact me whenever necessary. @@
@@ Thanks for you interests! @@

Pybullet-implementation of the multi-goal robotics environment originally from Open AI Gym.

This version uses a kuka iiwa14 7DoF arm, equipped with a robotiq85 two finger gripper or a simple parallel jaw.

The basic four tasks are basically the same as the ones in the OpenAI Gym: Reach, Push, Pick and Place, Slide. However, it may look funny when the robot picks up a block with the robotiq85 gripper, since it's under-actuated and thus makes it hard to fine-tune the simulation. You can use the parallel jaw gripper, which is effectively the same as the OpenAI one.

I have implemented some goal-conditioned RL algos in my another repo, using the original Gym environment. There are DDPG-HER, SAC-HER, and others. DRL_Implementation. Expired mujoco license got me here. I will also retrain those agents and pose performance ASAP.

This package also provides some harder tasks for long-horizon sparse reward robotic arm manipulation tasks on this package as well. All the environments have been summarised in a paper. The newest release is the most recommended. There are still on-going updates for this package, the [v1.0 release] was the version published on Taros 2021 and ArXiv. Due to further development, the description in this paper may not be exactly the same with the master branch.

The following tasks are supported in the v1.3 branch:

1. Reach, push, pick-and-place, slide as the gym-robotics tasks;
2. Four Multi-step tasks described in the Taros paper
3. Two shape-assemble tasks (block-fitting & reaching) with pushing primitive actions (continuous & discrete)
4. An insertion task with 6 DoF gripper frame control

@InProceedings{yang2021pmg,
author="Yang, Xintong and Ji, Ze and Wu, Jing and Lai, Yu-Kun",
title="An Open-Source Multi-goal Reinforcement Learning Environment for Robotic Manipulation with Pybullet",
booktitle="Towards Autonomous Robotic Systems",
year="2021",
publisher="Springer International Publishing",
pages="14--24",
isbn="978-3-030-89177-0"
}

Installation

git clone https://github.com/IanYangChina/pybullet_multigoal_gym.git
cd pybullet_multigoal_gym
pip install -r requirements.txt
pip install .

Some info

Observation, state, action, goal and reward are all setup to be the same as the original environment.

Observation is a dictionary, containing the state, desired goal, achieved goal and other sensory data.

Use the make_env(...) method to make your environments. Due to backend differences, the render() method should not need to be called by users. To run experiment headless, make environment with render=False. To run experiment with image observations, make environment with image_observation=True. Only the Reach, PickAndPlace and Push envs support image observation. Users can define their own camera for observation and goal images. The camera id -1 stands for a on-hand camera. See examples below.

Try it out

See the examples folder for more scripts to play with.

# Single-stage manipulation environments
# Reach, Push, PickAndPlace, Slide
import pybullet_multigoal_gym as pmg
# Install matplotlib if you want to use imshow to view the goal images
import matplotlib.pyplot as plt

camera_setup = [
    {
        'cameraEyePosition': [-0.9, -0.0, 0.4],
        'cameraTargetPosition': [-0.45, -0.0, 0.0],
        'cameraUpVector': [0, 0, 1],
        'render_width': 224,
        'render_height': 224
    },
    {
        'cameraEyePosition': [-1.0, -0.25, 0.6],
        'cameraTargetPosition': [-0.6, -0.05, 0.2],
        'cameraUpVector': [0, 0, 1],
        'render_width': 224,
        'render_height': 224
    },
]

env = pmg.make_env(
    # task args ['reach', 'push', 'slide', 'pick_and_place', 
    #            'block_stack', 'block_rearrange', 'chest_pick_and_place', 'chest_push']
    task='block_stack',
    gripper='parallel_jaw',
    num_block=4,  # only meaningful for multi-block tasks, up to 5 blocks
    render=True,
    binary_reward=True,
    max_episode_steps=50,
    # image observation args
    image_observation=True,
    depth_image=False,
    goal_image=True,
    visualize_target=True,
    camera_setup=camera_setup,
    observation_cam_id=[0],
    goal_cam_id=1)

f, axarr = plt.subplots(1, 2)
obs = env.reset()
while True:
    action = env.action_space.sample()
    obs, reward, done, info = env.step(action)
    print('state: ', obs['state'], '\n',
          'desired_goal: ', obs['desired_goal'], '\n',
          'achieved_goal: ', obs['achieved_goal'], '\n',
          'reward: ', reward, '\n')
    axarr[0].imshow(obs['desired_goal_img'])
    axarr[1].imshow(obs['achieved_goal_img'])
    plt.pause(0.00001)
    if done:
        env.reset()

Scenes

Update log

2020.12.26 --- Add hierarchical environments for a pick and place task, with image observation and goal supported. See the above example.

2020.12.28 --- Add image observation to non-hierarchical environments.

2021.01.14 --- Add parallel jaw gripper.

2021.03.08 --- Add a make_env(...) method to replace the pre-registration codes.

2021.03.09 --- Add multi-block stacking and re-arranging tasks

2021.03.12 --- Add multi-block tasks with a chest

2021.03.17 --- Joint space control support

2021.03.18 --- Finish curriculum; add on-hand camera observation

2021.11.11 --- Finish task decomposition, subgoal generation codes and some compatibility issues, new release

2021.12.03 --- Add shape assembly task with push primitive support (discrete & continuous)

2022.05.06 --- Clean up stuff; add insertion task; remove hierarchical env codes.