Trivial: MS-HAB to docs

haosulab · Dec 20, 2024 · f4a9607 · f4a9607
1 parent be920fd
commit f4a9607
Show file tree

Hide file tree

Showing 10 changed files with 93 additions and 8 deletions.
diff --git a/docs/generate_task_docs.py b/docs/generate_task_docs.py
@@ -31,19 +31,22 @@
 
     "mobile_manipulation": """# Mobile Manipulation Tasks
 
-These are tasks where a mobile manipulator is used to manipulate objects. This cateogry primarily uses robots with mobile bases like Fetch or Stretch robots""",
+These are tasks where a mobile manipulator is used to manipulate objects. This cateogry primarily uses robots with mobile bases like Fetch or Stretch robots.
+
+For additional tasks, including scene-level mobile manipulation, please check out the [community contributions page](../community_contributions/index.md).
+""",
 
     "quadruped": """# Quadruped Tasks
 
-These are tasks where a quadruped robot is used for locomotion and/or manipulation. This cateogry primarily uses robots with four legs like the ANYmal or Unitree go robots""",
+These are tasks where a quadruped robot is used for locomotion and/or manipulation. This cateogry primarily uses robots with four legs like the ANYmal or Unitree go robots.""",
 
     "control": """# Control Tasks
 
-These are classic control tasks where the objective is to control a robot to reach a particular state, similar to the [DM Control suite](https://github.com/deepmind/dm_control) but with GPU parallelized simulation and rendering""",
+These are classic control tasks where the objective is to control a robot to reach a particular state, similar to the [DM Control suite](https://github.com/deepmind/dm_control) but with GPU parallelized simulation and rendering.""",
 
     "drawing": """# Drawing Tasks
 
-These are tasks where the robot is controlled to draw a specific shape or pattern""",
+These are tasks where the robot is controlled to draw a specific shape or pattern.""",
 }
 import urllib.request
 import mani_skill.envs

diff --git a/docs/source/_static/env_thumbnails/ManiSkill-HAB_rt_poster.png b/docs/source/_static/env_thumbnails/ManiSkill-HAB_rt_poster.png
diff --git a/docs/source/_static/env_thumbnails/ManiSkill-HAB_rt_thumb.png b/docs/source/_static/env_thumbnails/ManiSkill-HAB_rt_thumb.png
diff --git a/docs/source/tasks/community_contributions/index.md b/docs/source/tasks/community_contributions/index.md
@@ -0,0 +1,78 @@
+<!-- THIS IS ALL GENERATED DOCUMENTATION. DO NOT MODIFY THIS FILE -->
+[asset-badge]: https://img.shields.io/badge/download%20asset-yes-blue.svg
+[dense-reward-badge]: https://img.shields.io/badge/dense%20reward-yes-green.svg
+[sparse-reward-badge]: https://img.shields.io/badge/sparse%20reward-yes-green.svg
+[no-dense-reward-badge]: https://img.shields.io/badge/dense%20reward-no-red.svg
+[no-sparse-reward-badge]: https://img.shields.io/badge/sparse%20reward-no-red.svg
+[demos-badge]: https://img.shields.io/badge/demos-yes-green.svg
+# Community Contributed Projects
+
+These are tasks and external libraries contributed by the community, built in ManiSkill.
+The document here has both a high-level overview/list of each project in a table as well as detailed task cards with video demonstrations after.
+
+## Task Table
+Table of all tasks/environments in this category. Task column is the environment ID, Preview is a thumbnail pair of the first and last frames of an example success demonstration. Max steps is the task's default max episode steps, generally tuned for RL workflows.
+<table class="table">
+<thead>
+<tr class="row-odd">
+<th class="head"><p>Task</p></th>
+<th class="head"><p>Preview</p></th>
+<th class="head"><p>Dense Reward</p></th>
+<th class="head"><p>Success/Fail Conditions</p></th>
+<th class="head"><p>Demos</p></th>
+<th class="head"><p>Max Episode Steps</p></th>
+</tr>
+</thead>
+<tbody>
+<tr class="row-odd">
+<td><p><a href="https://arth-shukla.github.io/mshab/" target="_blank">ManiSkill-HAB</a></p></td>
+<td><div style='display:flex;gap:4px;align-items:center'><img style='min-width:min(50%, 200px);max-width:200px;height:auto' src='../../_static/env_thumbnails/ManiSkill-HAB_rt_thumb.png' alt='ManiSkill-HAB'></td>
+<td><p>✅</p></td>
+<td><p>✅</p></td>
+<td><p>✅</p></td>
+<td><p>Full Task: 1200-2000<br /><br />Subtasks: 200</p></td>
+</tr>
+</tbody>
+</table>
+
+## ManiSkill-HAB
+
+**[Paper](https://arxiv.org/abs/2412.13211)** | **[Website](https://arth-shukla.github.io/mshab)** | **[Code](https://github.com/arth-shukla/mshab)** | **[Models](https://huggingface.co/arth-shukla/mshab_checkpoints)** | **[Dataset](https://arth-shukla.github.io/mshab/#dataset-section)**
+
+![dense-reward][dense-reward-badge]
+![sparse-reward][sparse-reward-badge]
+![demos][demos-badge]
+:::{dropdown} Task Card
+:icon: note
+:color: primary
+
+**Task Description:**
+
+ManiSkill-HAB contains 3 long-horizon tasks:
+- **TidyHouse**: Move 5 target objects to different open receptacles (e.g. table, counter, etc).
+- **PrepareGroceries**: Move 2 objects from the opened fridge to goal positions on the counter,
+then 1 object from the counter to the fridge.
+- **SetTable**: Move 1 bowl from the closed drawer to the dining table and 1 apple from the
+closed fridge to the same dining table.
+
+These tasks are split into 4 subtasks: Pick Place, Open, and Close.
+
+**Randomizations:**
+- Target object/articulation/goal (depending on subtask)
+- Base x/y position and z rotation, arm and torso qpos noise
+
+**Success Conditions:**
+- Pick: Object grasped, end-effector at rest position, robot static
+- Place: Object within 15cm of goal position, end-effector at rest position, robot static
+- Open: Articulation open, end-effector at rest position, robot static
+- Close: Articulation closed, end-effector at rest position, robot static
+
+**Fail Conditions:**
+- Robot exceeds cumulative collision force threshold (5000N for Pick, 7500N for Place, 10,000N for Open/Close)
+:::
+
+<div style="display: flex; justify-content: center;">
+<video preload="none" controls="True" width="100%" style="max-width: min(100%, 512px);" poster="../../_static/env_thumbnails/ManiSkill-HAB_rt_poster.png">
+<source src="https://github.com/haosulab/ManiSkill/raw/main/figures/environment_demos/ManiSkill-HAB_rt.mp4" type="video/mp4">
+</video>
+</div>
diff --git a/docs/source/tasks/control/index.md b/docs/source/tasks/control/index.md
@@ -7,7 +7,7 @@
 [demos-badge]: https://img.shields.io/badge/demos-yes-green.svg
 # Control Tasks
 
-These are classic control tasks where the objective is to control a robot to reach a particular state, similar to the [DM Control suite](https://github.com/deepmind/dm_control) but with GPU parallelized simulation and rendering
+These are classic control tasks where the objective is to control a robot to reach a particular state, similar to the [DM Control suite](https://github.com/deepmind/dm_control) but with GPU parallelized simulation and rendering.
 The document here has both a high-level overview/list of all tasks in a table as well as detailed task cards with video demonstrations after.
 
 ## Task Table

diff --git a/docs/source/tasks/drawing/index.md b/docs/source/tasks/drawing/index.md
@@ -7,7 +7,7 @@
 [demos-badge]: https://img.shields.io/badge/demos-yes-green.svg
 # Drawing Tasks
 
-These are tasks where the robot is controlled to draw a specific shape or pattern
+These are tasks where the robot is controlled to draw a specific shape or pattern.
 The document here has both a high-level overview/list of all tasks in a table as well as detailed task cards with video demonstrations after.
 
 ## Task Table

diff --git a/docs/source/tasks/index.md b/docs/source/tasks/index.md
@@ -38,4 +38,5 @@ mobile_manipulation/index
 dextrous/index
 digital_twins/index
 drawing/index
+community_contributions/index
 ```
diff --git a/docs/source/tasks/mobile_manipulation/index.md b/docs/source/tasks/mobile_manipulation/index.md
@@ -7,7 +7,10 @@
 [demos-badge]: https://img.shields.io/badge/demos-yes-green.svg
 # Mobile Manipulation Tasks
 
-These are tasks where a mobile manipulator is used to manipulate objects. This cateogry primarily uses robots with mobile bases like Fetch or Stretch robots
+These are tasks where a mobile manipulator is used to manipulate objects. This cateogry primarily uses robots with mobile bases like Fetch or Stretch robots.
+
+For additional tasks, including scene-level mobile manipulation, please check out the [community contributions page](../community_contributions/index.md).
+
 The document here has both a high-level overview/list of all tasks in a table as well as detailed task cards with video demonstrations after.
 
 ## Task Table

diff --git a/docs/source/tasks/quadruped/index.md b/docs/source/tasks/quadruped/index.md
@@ -7,7 +7,7 @@
 [demos-badge]: https://img.shields.io/badge/demos-yes-green.svg
 # Quadruped Tasks
 
-These are tasks where a quadruped robot is used for locomotion and/or manipulation. This cateogry primarily uses robots with four legs like the ANYmal or Unitree go robots
+These are tasks where a quadruped robot is used for locomotion and/or manipulation. This cateogry primarily uses robots with four legs like the ANYmal or Unitree go robots.
 The document here has both a high-level overview/list of all tasks in a table as well as detailed task cards with video demonstrations after.
 
 ## Task Table

diff --git a/figures/environment_demos/ManiSkill-HAB_rt.mp4 b/figures/environment_demos/ManiSkill-HAB_rt.mp4