gym-aloha

A gym environment for ALOHA

Installation

Create a virtual environment with Python 3.10 and activate it, e.g. with miniconda:

conda create -y -n aloha python=3.10 && conda activate aloha

Install gym-aloha:

pip install gym-aloha

Quickstart

# example.py
import imageio
import gymnasium as gym
import numpy as np
import gym_aloha

env = gym.make("gym_aloha/AlohaInsertion-v0")
observation, info = env.reset()
frames = []

for _ in range(1000):
    action = env.action_space.sample()
    observation, reward, terminated, truncated, info = env.step(action)
    image = env.render()
    frames.append(image)

    if terminated or truncated:
        observation, info = env.reset()

env.close()
imageio.mimsave("example.mp4", np.stack(frames), fps=25)

Description

Aloha environment.

Two tasks are available:

• TransferCubeTask: The right arm needs to first pick up the red cube lying on the table, then place it inside the gripper of the other arm.
• InsertionTask: The left and right arms need to pick up the socket and peg respectively, and then insert in mid-air so the peg touches the “pins” inside the socket.

Action Space

The action space consists of continuous values for each arm and gripper, resulting in a 14-dimensional vector:

• Six values for each arm's joint positions (absolute values).
• One value for each gripper's position, normalized between 0 (closed) and 1 (open).

Observation Space

Observations are provided as a dictionary with the following keys:

• qpos and qvel: Position and velocity data for the arms and grippers.
• images: Camera feeds from different angles.
• env_state: Additional environment state information, such as positions of the peg and sockets.

Rewards

• TransferCubeTask:
  • 1 point for holding the box with the right gripper.
  • 2 points if the box is lifted with the right gripper.
  • 3 points for transferring the box to the left gripper.
  • 4 points for a successful transfer without touching the table.
• InsertionTask:
  • 1 point for touching both the peg and a socket with the grippers.
  • 2 points for grasping both without dropping them.
  • 3 points if the peg is aligned with and touching the socket.
  • 4 points for successful insertion of the peg into the socket.

Success Criteria

Achieving the maximum reward of 4 points.

Starting State

The arms and the items (block, peg, socket) start at a random position and angle.

Arguments

>>> import gymnasium as gym
>>> import gym_aloha
>>> env = gym.make("gym_aloha/AlohaInsertion-v0", obs_type="pixels", render_mode="rgb_array")
>>> env
<TimeLimit<OrderEnforcing<PassiveEnvChecker<AlohaEnv<gym_aloha/AlohaInsertion-v0>>>>>

• obs_type: (str) The observation type. Can be either pixels or pixels_agent_pos. Default is pixels.

• render_mode: (str) The rendering mode. Only rgb_array is supported for now.

• observation_width: (int) The width of the observed image. Default is 640.

• observation_height: (int) The height of the observed image. Default is 480.

• visualization_width: (int) The width of the visualized image. Default is 640.

• visualization_height: (int) The height of the visualized image. Default is 480.

🔧 GPU Rendering (EGL)

Rendering on the GPU can be significantly faster than CPU. However, MuJoCo may silently fall back to CPU rendering if EGL is not properly configured. To force GPU rendering and avoid fallback issues, you can use the following snippet:

import distutils.util
import os
import subprocess

if subprocess.run('nvidia-smi').returncode:
  raise RuntimeError(
      'Cannot communicate with GPU. '
      'Make sure you are using a GPU runtime. '
      'Go to the Runtime menu and select Choose runtime type.'
  )

# Add an ICD config so that glvnd can pick up the Nvidia EGL driver.
# This is usually installed as part of an Nvidia driver package, but the
# kernel doesn't install its driver via APT, and as a result the ICD is missing.
# (https://github.com/NVIDIA/libglvnd/blob/master/src/EGL/icd_enumeration.md)
NVIDIA_ICD_CONFIG_PATH = '/usr/share/glvnd/egl_vendor.d/10_nvidia.json'
if not os.path.exists(NVIDIA_ICD_CONFIG_PATH):
  with open(NVIDIA_ICD_CONFIG_PATH, 'w') as f:
    f.write("""{
    "file_format_version" : "1.0.0",
    "ICD" : {
        "library_path" : "libEGL_nvidia.so.0"
    }
}
""")

# Check if installation was successful.
try:
  print('Checking that the installation succeeded:')
  import mujoco
  from mujoco import rollout
  mujoco.MjModel.from_xml_string('<mujoco/>')
except Exception as e:
  raise e from RuntimeError(
      'Something went wrong during installation. Check the shell output above '
      'for more information.\n'
      'If using a hosted Colab runtime, make sure you enable GPU acceleration '
      'by going to the Runtime menu and selecting "Choose runtime type".')

print('Installation successful.')

# Tell XLA to use Triton GEMM, this improves steps/sec by ~30% on some GPUs
xla_flags = os.environ.get('XLA_FLAGS', '')
xla_flags += ' --xla_gpu_triton_gemm_any=True'
os.environ['XLA_FLAGS'] = xla_flags

Contribute

Instead of using pip directly, we use poetry for development purposes to easily track our dependencies. If you don't have it already, follow the instructions to install it.

Install the project with dev dependencies:

poetry install --all-extras

Follow our style

# install pre-commit hooks
pre-commit install

# apply style and linter checks on staged files
pre-commit

Acknowledgment

gym-aloha is adapted from ALOHA