Release Notes#

v1.1.0: MO-Gymnasium 1.1.0 Release: New MuJoCo environments, Mirrored Deep Sea Treasure, Fruit Tree rendering, and more#

Released on 2024-03-11 - GitHub - PyPI

Environments

Other improvements and utils

Documentation

Bug fixes

Full Changelog: v1.0.1...v1.1.0

v1.0.1: MO-Gymnasium 1.0.1 Release: Support Gymnasium 0.29, breakable-bottles pygame render, and more#

Released on 2023-08-24 - GitHub - PyPI

Environments

Wrapper

Other improvements and utils

  • Modify LinearReward to return reward weights as part of info_dict by @ianleongudri in #69
  • Add warning for order of wrapping in the MORecordEpisodeStatistics Wrapper by @ffelten in #70
  • Support Gymnasium 0.29 by @LucasAlegre in #73

Documentation

Bug fixes

Full Changelog: v1.0.0...v1.0.1

v1.0.0: MO-Gymnasium becomes mature#

Released on 2023-06-12 - GitHub - PyPI

MO-Gymnasium 1.0.0 Release Notes

We are thrilled to introduce the mature release of MO-Gymnasium, a standardized API and collection of environments designed for Multi-Objective Reinforcement Learning (MORL).

MORL expands the capabilities of RL to scenarios where agents need to optimize multiple objectives, which may potentially conflict with each other. Each objective is represented by a distinct reward function. In this context, the agent learns to make trade-offs between these objectives based on a reward vector received after each step. For instance, in the well-known Mujoco halfcheetah environment, reward components are combined linearly using predefined weights as shown in the following code snippet from Gymnasium:

ctrl_cost = self.control_cost(action)
forward_reward = self._forward_reward_weight * x_velocity
reward = forward_reward - ctrl_cost

With MORL, users have the flexibility to determine the compromises they desire based on their preferences for each objective. Consequently, the environments in MO-Gymnasium do not have predefined weights. Thus, MO-Gymnasium extends the capabilities of Gymnasium to the multi-objective setting, where the agents receives a vectorial reward.

For example, here is an illustration of the multiple policies learned by an MORL agent for the mo-halfcheetah domain, balancing between saving battery and speed:

This release marks the first mature version of MO-Gymnasium within Farama, indicating that the API is stable, and we have achieved a high level of quality in this library.

API

import gymnasium as gym
import mo_gymnasium as mo_gym
import numpy as np

# It follows the original Gymnasium API ...
env = mo_gym.make('minecart-v0')

obs, info = env.reset()
# but vector_reward is a numpy array!
next_obs, vector_reward, terminated, truncated, info = env.step(your_agent.act(obs))

# Optionally, you can scalarize the reward function with the LinearReward wrapper.
# This allows to fall back to single objective RL
env = mo_gym.LinearReward(env, weight=np.array([0.8, 0.2, 0.2]))

Environments

We support environments ranging from MORL literature to inherently multi-objective problems in the RL literature such as Mujoco. An exhaustive list of environments is available on our documentation website.

Wrappers

Additionally, we provide a set of wrappers tailor made for MORL, such as MONormalizeReward which normalizes an element of the reward vector, or LinearWrapper which transforms the MOMDP into an MDP. See also our documentation.

New features and improvements

  • Bump highway-env version in #50
  • Add mo-lunar-lander-continuous-v2 and mo-hopper-2d-v4 environments in #51
  • Add normalized action option to water-reservoir-v0 in #52
  • Accept zero-dimension numpy array as discrete action in #55
  • Update pre-commit versions and fix small spelling mistake in #56
  • Add method to compute known Pareto Front of fruit tree in #57
  • Improve reward bounds on: Mario, minecart, mountain car, resource gathering, reacher in #59, #60, #61
  • Add Python 3.11 support, drop Python 3.7 in #65

Bug fixes and documentation updates

  • Fix water-reservoir bug caused by numpy randint deprecation in #53
  • Fix missing edit button in website in #58
  • Fix reward space and add reward bound tests in #62
  • Add MO-Gymnasium logo to docs in #64

Full Changelog: v0.3.4...v1.0.0

v0.3.4: MO-Gymnasium 0.3.4 Release: Known Pareto Front, improved renders and documentation#

Released on 2023-03-14 - GitHub - PyPI

Changelogs

Environments

  • Add new pixel art rendering for deep-sea-treasure-v0, resource-gathering-v0 and water-reservoir-v0 by @LucasAlegre in #41
  • Add pareto_front function to get known optimal front in DST, Minecart and Resource Gathering by @LucasAlegre and @ffelten in #45, #43;
  • Add deep-sea-treasure-concave-v0 by @ffelten in #43

Utils

Documentation

  • Improve documentation and README by @LucasAlegre in #40
  • Create docs/README.md to link to a new CONTRIBUTING.md for docs by @mgoulao in #42
  • Enable documentation versioning and release notes in website by @mgoulao in #46

New Contributors

Full Changelog: v0.3.3...0.3.4

v0.3.3: MO-Gymnasium 0.3.3 Release: Policy Evaluation bug fix, better documentation page#

Released on 2023-02-13 - GitHub - PyPI

New improvements/features

Bugs fixed

  • Fix highway env observation conversion by @LucasAlegre in #33
  • Fix bug in eval_mo which was passing None to all weight vectors
  • Fix minecart and water-reservoir ObservationSpace dtype and bounds

Documentation

Full Changelog: 0.3.2...v0.3.3

MO-Gymnasium 0.3.2 Release: Bug fixes, improved webpage#

Released on 2023-02-03 - GitHub - PyPI

Bug fixes

  • Bump highway-env version, to fix rendering
  • Add assets to the pypi release package

Documentation

  • Add gifs to the webpage

Full Changelog: 0.3.1...0.3.2

MO-Gymnasium 0.3.1 Release: Improved documentation and MuJoco MO-Reacher environment#

Released on 2023-02-02 - GitHub - PyPI

This minor release adds "mo-reacher-v4", a MuJoco version of the Reacher environment, fixes a bug in Lunar Lander and improves the library documentation.

Environments

Documentation

Bug Fixes

  • Hotfix lunar lander by @ffelten in #27
  • MORecordEpisodeStatistics return scalars when not VecEnv by @ffelten in #30

Full Changelog: 0.3.0...0.3.1

MO-Gymnasium 0.3.0 Release: Migrate to Gymnasium#

Released on 2023-01-23 - GitHub - PyPI

This release marks our first release as part of the Farama Foundation. Benefitting the Farama structure, this library should reach a higher level of quality and more integration with the tools from the RL community.

Breaking changes

  • The package has been renamed MO-Gymnasium (it was previously called MO-Gym). We now also rely on Gymnasium instead of Gym, see the by @LucasAlegre in #16
  • Environments are now under the envs package, was previously accessible from root e.g.mo_gymnasium.deep_sea_treasure -> mo_gymnasium.envs.deep_sea_treasure.

Environments

Quality of life

Community

Full Changelog: 0.2.1...0.3.0

0.2.1#

Released on 2022-12-09 - GitHub - PyPI

  • 5 new environments: fishwood-v0 (ESR), mo-MountainCarContinuous-v0, water-reservoir-v0, mo-highway-v0 and mo-highway-fast-v0;
  • Revamped README file;
  • Linting and automatic imports optimization;
  • Updated bib file and citation;
  • Few bugfixes.

0.2.0#

Released on 2022-09-25 - GitHub - PyPI

Support for new Gym>=0.26 API

0.1.2#

Released on 2022-09-25 - GitHub - PyPI

0.1.1#

Released on 2022-08-24 - GitHub - PyPI