| Name | gym-pusht JSON |
| Version |
0.1.5
JSON |
| download |
| home_page | None |
| Summary | A gymnasium environment for PushT. |
| upload_time | 2024-07-05 10:13:17 |
| maintainer | None |
| docs_url | None |
| author | Rémi Cadène |
| requires_python | <4.0,>=3.10 |
| license | Apache-2.0 |
| keywords |
|
| VCS |
|
| bugtrack_url |
|
| requirements |
No requirements were recorded.
|
| Travis-CI |
No Travis.
|
| coveralls test coverage |
No coveralls.
|
# gym-pusht
A gymnasium environment PushT.
<img src="http://remicadene.com/assets/gif/pusht_diffusion.gif" width="50%" alt="Diffusion policy on PushT env"/>
## Installation
Create a virtual environment with Python 3.10 and activate it, e.g. with [`miniconda`](https://docs.anaconda.com/free/miniconda/index.html):
```bash
conda create -y -n pusht python=3.10 && conda activate pusht
```
Install gym-pusht:
```bash
pip install gym-pusht
```
## Quick start
```python
# example.py
import gymnasium as gym
import gym_pusht
env = gym.make("gym_pusht/PushT-v0", render_mode="human")
observation, info = env.reset()
for _ in range(1000):
action = env.action_space.sample()
observation, reward, terminated, truncated, info = env.step(action)
image = env.render()
if terminated or truncated:
observation, info = env.reset()
env.close()
```
## Description
PushT environment.
The goal of the agent is to push the block to the goal zone. The agent is a circle and the block is a tee shape.
### Action Space
The action space is continuous and consists of two values: [x, y]. The values are in the range [0, 512] and
represent the target position of the agent.
### Observation Space
If `obs_type` is set to `state`, the observation space is a 5-dimensional vector representing the state of the
environment: [agent_x, agent_y, block_x, block_y, block_angle]. The values are in the range [0, 512] for the agent
and block positions and [0, 2*pi] for the block angle.
If `obs_type` is set to `environment_state_agent_pos` the observation space is a dictionary with:
- `environment_state`: 16-dimensional vector representing the keypoint locations of the T (in [x0, y0, x1, y1, ...]
format). The values are in the range [0, 512].
- `agent_pos`: A 2-dimensional vector representing the position of the robot end-effector.
If `obs_type` is set to `pixels`, the observation space is a 96x96 RGB image of the environment.
### Rewards
The reward is the coverage of the block in the goal zone. The reward is 1.0 if the block is fully in the goal zone.
### Success Criteria
The environment is considered solved if the block is at least 95% in the goal zone.
### Starting State
The agent starts at a random position and the block starts at a random position and angle.
### Episode Termination
The episode terminates when the block is at least 95% in the goal zone.
### Arguments
```python
>>> import gymnasium as gym
>>> import gym_pusht
>>> env = gym.make("gym_pusht/PushT-v0", obs_type="state", render_mode="rgb_array")
>>> env
<TimeLimit<OrderEnforcing<PassiveEnvChecker<PushTEnv<gym_pusht/PushT-v0>>>>>
```
* `obs_type`: (str) The observation type. Can be either `state`, `environment_state_agent_pos`, `pixels` or `pixels_agent_pos`. Default is `state`.
* `block_cog`: (tuple) The center of gravity of the block if different from the center of mass. Default is `None`.
* `damping`: (float) The damping factor of the environment if different from 0. Default is `None`.
* `render_mode`: (str) The rendering mode. Can be either `human` or `rgb_array`. Default is `rgb_array`.
* `observation_width`: (int) The width of the observed image. Default is `96`.
* `observation_height`: (int) The height of the observed image. Default is `96`.
* `visualization_width`: (int) The width of the visualized image. Default is `680`.
* `visualization_height`: (int) The height of the visualized image. Default is `680`.
### Reset Arguments
Passing the option `options["reset_to_state"]` will reset the environment to a specific state.
> [!WARNING]
> For legacy compatibility, the inner functioning has been preserved, and the state set is not the same as the
> the one passed in the argument.
```python
>>> import gymnasium as gym
>>> import gym_pusht
>>> env = gym.make("gym_pusht/PushT-v0")
>>> state, _ = env.reset(options={"reset_to_state": [0.0, 10.0, 20.0, 30.0, 1.0]})
>>> state
array([ 0. , 10. , 57.866196, 50.686398, 1. ],
dtype=float32)
```
## Version History
* v0: Original version
## References
* TODO:
## 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](https://python-poetry.org/docs/#installation) to install it.
Install the project with dev dependencies:
```bash
poetry install --all-extras
```
### Follow our style
```bash
# install pre-commit hooks
pre-commit install
# apply style and linter checks on staged files
pre-commit
```
## Acknowledgment
gym-pusht is adapted from [Diffusion Policy](https://diffusion-policy.cs.columbia.edu/)
Raw data
{
"_id": null,
"home_page": null,
"name": "gym-pusht",
"maintainer": null,
"docs_url": null,
"requires_python": "<4.0,>=3.10",
"maintainer_email": null,
"keywords": null,
"author": "R\u00e9mi Cad\u00e8ne",
"author_email": "re.cadene@gmail.com",
"download_url": "https://files.pythonhosted.org/packages/57/ef/45056f81ba2a7567289ce5f9e1806654ab4d4df469ee527a52637b81834a/gym_pusht-0.1.5.tar.gz",
"platform": null,
"description": "# gym-pusht\n\nA gymnasium environment PushT.\n\n<img src=\"http://remicadene.com/assets/gif/pusht_diffusion.gif\" width=\"50%\" alt=\"Diffusion policy on PushT env\"/>\n\n\n## Installation\n\nCreate a virtual environment with Python 3.10 and activate it, e.g. with [`miniconda`](https://docs.anaconda.com/free/miniconda/index.html):\n```bash\nconda create -y -n pusht python=3.10 && conda activate pusht\n```\n\nInstall gym-pusht:\n```bash\npip install gym-pusht\n```\n\n\n## Quick start\n\n```python\n# example.py\nimport gymnasium as gym\nimport gym_pusht\n\nenv = gym.make(\"gym_pusht/PushT-v0\", render_mode=\"human\")\nobservation, info = env.reset()\n\nfor _ in range(1000):\n action = env.action_space.sample()\n observation, reward, terminated, truncated, info = env.step(action)\n image = env.render()\n\n if terminated or truncated:\n observation, info = env.reset()\n\nenv.close()\n```\n\n## Description\n\nPushT environment.\n\nThe goal of the agent is to push the block to the goal zone. The agent is a circle and the block is a tee shape.\n\n### Action Space\n\nThe action space is continuous and consists of two values: [x, y]. The values are in the range [0, 512] and\nrepresent the target position of the agent.\n\n### Observation Space\n\nIf `obs_type` is set to `state`, the observation space is a 5-dimensional vector representing the state of the\nenvironment: [agent_x, agent_y, block_x, block_y, block_angle]. The values are in the range [0, 512] for the agent\nand block positions and [0, 2*pi] for the block angle.\n\nIf `obs_type` is set to `environment_state_agent_pos` the observation space is a dictionary with:\n - `environment_state`: 16-dimensional vector representing the keypoint locations of the T (in [x0, y0, x1, y1, ...]\n format). The values are in the range [0, 512].\n - `agent_pos`: A 2-dimensional vector representing the position of the robot end-effector.\n\nIf `obs_type` is set to `pixels`, the observation space is a 96x96 RGB image of the environment.\n\n### Rewards\n\nThe reward is the coverage of the block in the goal zone. The reward is 1.0 if the block is fully in the goal zone.\n\n### Success Criteria\n\nThe environment is considered solved if the block is at least 95% in the goal zone.\n\n### Starting State\n\nThe agent starts at a random position and the block starts at a random position and angle.\n\n### Episode Termination\n\nThe episode terminates when the block is at least 95% in the goal zone.\n\n### Arguments\n\n```python\n>>> import gymnasium as gym\n>>> import gym_pusht\n>>> env = gym.make(\"gym_pusht/PushT-v0\", obs_type=\"state\", render_mode=\"rgb_array\")\n>>> env\n<TimeLimit<OrderEnforcing<PassiveEnvChecker<PushTEnv<gym_pusht/PushT-v0>>>>>\n```\n\n* `obs_type`: (str) The observation type. Can be either `state`, `environment_state_agent_pos`, `pixels` or `pixels_agent_pos`. Default is `state`.\n\n* `block_cog`: (tuple) The center of gravity of the block if different from the center of mass. Default is `None`.\n\n* `damping`: (float) The damping factor of the environment if different from 0. Default is `None`.\n\n* `render_mode`: (str) The rendering mode. Can be either `human` or `rgb_array`. Default is `rgb_array`.\n\n* `observation_width`: (int) The width of the observed image. Default is `96`.\n\n* `observation_height`: (int) The height of the observed image. Default is `96`.\n\n* `visualization_width`: (int) The width of the visualized image. Default is `680`.\n\n* `visualization_height`: (int) The height of the visualized image. Default is `680`.\n\n### Reset Arguments\n\nPassing the option `options[\"reset_to_state\"]` will reset the environment to a specific state.\n\n> [!WARNING]\n> For legacy compatibility, the inner functioning has been preserved, and the state set is not the same as the\n> the one passed in the argument.\n\n```python\n>>> import gymnasium as gym\n>>> import gym_pusht\n>>> env = gym.make(\"gym_pusht/PushT-v0\")\n>>> state, _ = env.reset(options={\"reset_to_state\": [0.0, 10.0, 20.0, 30.0, 1.0]})\n>>> state\narray([ 0. , 10. , 57.866196, 50.686398, 1. ],\n dtype=float32)\n```\n\n\n## Version History\n\n* v0: Original version\n\n\n## References\n\n* TODO:\n\n\n## Contribute\n\nInstead of using `pip` directly, we use `poetry` for development purposes to easily track our dependencies.\nIf you don't have it already, follow the [instructions](https://python-poetry.org/docs/#installation) to install it.\n\nInstall the project with dev dependencies:\n```bash\npoetry install --all-extras\n```\n\n### Follow our style\n\n```bash\n# install pre-commit hooks\npre-commit install\n\n# apply style and linter checks on staged files\npre-commit\n```\n\n## Acknowledgment\n\ngym-pusht is adapted from [Diffusion Policy](https://diffusion-policy.cs.columbia.edu/)\n",
"bugtrack_url": null,
"license": "Apache-2.0",
"summary": "A gymnasium environment for PushT.",
"version": "0.1.5",
"project_urls": null,
"split_keywords": [],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "a4cf6b92e0b42b14e0aef108ab56c77a3caad661008bd109d6f856d6c2f8acca",
"md5": "226f4f60d818fdd21368106b4646044a",
"sha256": "d9e3ba5f44916dc4a802d71764b08f4e7e09bda256e25af9dda16e9364dc777f"
},
"downloads": -1,
"filename": "gym_pusht-0.1.5-py3-none-any.whl",
"has_sig": false,
"md5_digest": "226f4f60d818fdd21368106b4646044a",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": "<4.0,>=3.10",
"size": 16873,
"upload_time": "2024-07-05T10:13:16",
"upload_time_iso_8601": "2024-07-05T10:13:16.221028Z",
"url": "https://files.pythonhosted.org/packages/a4/cf/6b92e0b42b14e0aef108ab56c77a3caad661008bd109d6f856d6c2f8acca/gym_pusht-0.1.5-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "57ef45056f81ba2a7567289ce5f9e1806654ab4d4df469ee527a52637b81834a",
"md5": "3afa2eb93b5d8c9aec6d0ddcf05dc699",
"sha256": "981e135f6e0ca91e4ec63603e9551bc77cba989d06a2888ed31a1d68f7cbdae2"
},
"downloads": -1,
"filename": "gym_pusht-0.1.5.tar.gz",
"has_sig": false,
"md5_digest": "3afa2eb93b5d8c9aec6d0ddcf05dc699",
"packagetype": "sdist",
"python_version": "source",
"requires_python": "<4.0,>=3.10",
"size": 16702,
"upload_time": "2024-07-05T10:13:17",
"upload_time_iso_8601": "2024-07-05T10:13:17.415072Z",
"url": "https://files.pythonhosted.org/packages/57/ef/45056f81ba2a7567289ce5f9e1806654ab4d4df469ee527a52637b81834a/gym_pusht-0.1.5.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-07-05 10:13:17",
"github": false,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"lcname": "gym-pusht"
}
