ManiSkill
SoftwareManiSkill supplies robotics simulation tasks, environment code and data-collection/learning workflows. [1] [2] [3] [4] [5]
Installation & first result →Official websiteGetting started
- For
- Robotics developers selecting a concrete sensor, interface or simulation workflow [1] [2] [3] [4] [5]
- Source rights
- Named licenses are documented by part. Review licenses
Source researched · Not tested by us
Sources checked · Research details
Licenses & related hardware
Overview
ManiSkill supplies robotics simulation tasks, environment code and data-collection/learning workflows. The selected PickCube implementation provides a concrete task to inspect. Its simulation and rendering facilities are different from physical robot execution, and upstream throughput reports are not benchmarks reproduced by this catalog. [1] [4]
Is this for you?
What you need
- Operating system
- Linux/NVIDIA: documented CPU/GPU simulation and rendering
- Windows/NVIDIA or Windows/AMD: documented CPU simulation/rendering
- WSL: documented CPU simulation only
- macOS: conflicting guidance requires version-specific reconciliation
Upstream documented requirement/configuration only, not independently tested installation or compatibility.
- Dependencies
- Matchedmani_skill/PyTorch environment; SAPIEN-basedtasks; Vulkan driver forrendering
- Task-specificexternalassets and MS_ASSET_DIR configuration
Installation & first result
Open the upstream getting-started guide [1] [2] [3] [4] [5]
- Select the named hardware/model and corresponding source configuration before installation.
- Follow the cited starting route while keeping package, device and firmware versions matched; no operation reproduced here.
Editorial assessment of cited source; no runtime or hardware test.
- Installation
Documented mani_skill and PyTorch installation, separate nightly option; random-action PickCube-v1 is a documented first check, not executed here. Rendering requires a GPU/Vulkan driver.
Upstream documented requirement/configuration only, not independently tested installation or compatibility.
Project notes
Choose one task and the required execution path
The installation guide documents mani_skill and PyTorch package installation, with a separate nightly distribution. An unpinned package today is not necessarily this reviewed source. Its bounded installation check is the random-action example for PickCube-v1, limited in the guide to a short episode. No package, environment or random action was executed here. [5]
The guide distinguishes state-based simulation from rendering, which requires a GPU with a Vulkan driver. The supplied tables list Linux/NVIDIA for CPU simulation, GPU simulation and rendering; Windows/NVIDIA and Windows/AMD for CPU simulation and rendering; and WSL for CPU simulation only. Do not equate an available CPU simulation path or a rendered window with GPU-parallel simulation support on another platform. The macOS guidance conflicts: the installation prose says no support, while its table lists CPU simulation/rendering and it links macOS instructions; the README describes limited support. Reconcile version-specific macOS documentation before selecting that host. These are upstream statements, not operation independently verified here. [1] [5]
Assets and demonstrations are not all downloaded by default. The guide uses MS_ASSET_DIR to select their location and a separate environment variable to skip an asset-download prompt. A source checkout is therefore not proof that every chosen robot/object task has its data present. Inspect the named task and required asset terms before downloading or redistributing them. [4] [5]
Keep code and model assets separate
The root code notice is Apache-2.0, with separate third-party notices. The README describes assets as CC-BY-NC-4.0; those restrictions are not replaced by the code license. No external asset payload, demonstration dataset, scene loading, training, GPU throughput or sim-to-real behavior was independently validated. [1] [2] [3]
What we checked
Pinned documentation/selected source review only; no build, installation, live device/network, simulator, calibration or performance test.
Not yet established
- Requirements: host computer
- Version-specific hardware compatibility (ecosystem tags alone do not establish it)
Updates
- Upstream last updated
- Not established
- Latest release
- Not established
- Profile updated
- Nov 1, 2026
- Sources last checked
- Nov 1, 2026
- Next review
- Dec 1, 2026
- Reviewed revision
62ff3a5
Source files and licenses
Distinct robot-facing source implementation/interface; no purchased hardware design or untested performance claim.
| Layer | Availability / license |
|---|---|
| software | available · Apache-2.0 code; assets separately restricted Scoped source declaration; purchased hardware, firmware, binaries, assets and external dependencies require separate terms. [1] [2] [3] [4] [5] |
| documentation | available · Apache-2.0 code; assets separately restricted Scoped source declaration; purchased hardware, firmware, binaries, assets and external dependencies require separate terms. [1] [2] [3] [4] [5] |
No separate physical design, pretrained distribution or dataset is selected as the released contribution in this scoped software review.
References
Each source links to the exact revision we read. “Used for” shows where this page relies on it.
- S001 README.md
Revision, date & where cited
mani-skill/ManiSkill contributors · github.com · revision
62ff3a5· README.md at · Accessed Nov 1, 2026
Used for: Overview, Is this for you?, What you need, Installation & first result and 3 more - S002 LICENSE
Revision, date & where cited
mani-skill/ManiSkill contributors · github.com · revision
62ff3a5· LICENSE at · Accessed Nov 1, 2026
Used for: Is this for you?, What you need, Installation & first result, Keep code and model assets separate and 1 more - S003 LICENSE-3RD-PARTY
Revision, date & where cited
mani-skill/ManiSkill contributors · github.com · revision
62ff3a5· LICENSE-3RD-PARTY at · Accessed Nov 1, 2026
Used for: Is this for you?, What you need, Installation & first result, Keep code and model assets separate and 1 more - S004 mani_skill/envs/tasks/tabletop/pick_cube.py
Revision, date & where cited
mani-skill/ManiSkill contributors · github.com · revision
62ff3a5· mani_skill/envs/tasks/tabletop/pick_cube.py at · Accessed Nov 1, 2026
Used for: Overview, Is this for you?, What you need, Installation & first result and 2 more - S005 docs/source/user_guide/getting_started/installation.md
Revision, date & where cited
mani-skill/ManiSkill contributors · github.com · revision
62ff3a5· docs/source/user_guide/getting_started/installation.md at · Accessed Nov 1, 2026
Used for: Is this for you?, What you need, Installation & first result, Choose one task and the required execution path and 1 more
Correction guidance · Discussion is not enabled · Profile format 0.7.0-draft.5