Open Source RoboticsRobots, software & learning

LEAP Hand Isaac Lab reorientation

Software

Isaac Lab environment and deployment implementation for one-axis in-hand reorientation using LEAP Hand motor proprioception. [1] [3]

Installation & first result →Official websiteGetting started

For
Researchers with LEAP Hand and Isaac Lab experience [1]
Source rights
Named licenses are documented by part. Review licenses
Unknowns
operating system, compute requirements not fully established. Review gaps

Source researched · Not tested by us
Sources checked · Research details

Licenses & related hardware

Requirements · Research details and dates

On this page

Overview

This LEAP Hand extension provides an Isaac Lab environment for one-axis in-hand reorientation, with training and sim-to-real deployment code. The documented task uses motor proprioception rather than an external vision system. It is a distinct software workflow associated with the LEAP Hand hardware, not a claim of general-purpose dexterous manipulation. [1]

Is this for you?

Known limitations
  • Upstream real-world task uses a7.5cm PLA cube weighing281g; no general-object success is implied.
  • The README MIT declaration does not automatically settle Isaac dependencies, external SDKs or checkpoint-specific rights.
  • No simulator, trained policy, USB transport or physical hand was run.

Editorial scope assessment from the cited implementation/documentation; no execution or compatibility test.

[1] [2] [3] [4] [5] · Editorial assessment

What you need

Dependencies
  • README tested environment: Ubuntu22.04, IsaacSim4.5, IsaacLab2.1.0, Python3.10
  • rl_games for the documented training route
  • LEAP Hand Python SDK or ROS2 Humble SDK for the corresponding deployment route

[1] [2] [3] [4] [5]

Installation & first result

Open the upstream getting-started guide [1]

  • Match the documented simulator versions and editable extension install; begin with environment configuration/training inspection before selecting a policy and hardware deployment.
  • Preserve exact source/dependency/model versions and evaluate the selected workflow before connecting physical hardware or reusing data.

Editorial scope assessment from the cited implementation/documentation; no execution or compatibility test.

[1] [2] [3] [4] [5] · Editorial assessment

Installation

Match the documented simulator versions and editable extension install; begin with environment configuration/training inspection before selecting a policy and hardware deployment.

Editorial scope assessment from the cited implementation/documentation; no execution or compatibility test.

[1] [2] [3] [4] [5] · Editorial assessment

Hardware, models & data

Inputs and outputs

Simulated hand/cube observations to trained policy actions; optional physical-hand deployment code.

Editorial scope assessment from the cited implementation/documentation; no execution or compatibility test.

[1] [3] · Editorial assessment

Project notes

Keep the tested upstream stack together

The README reports Ubuntu 22.04, Isaac Sim 4.5, Isaac Lab 2.1.0 and a Python 3.10 conda environment. It describes an editable installation of source/LEAP_Isaaclab and a training/play route through rl_games using the Isaac-Reorient-Cube-Leap task. These are the upstream's documented versions and commands; they were not installed or reproduced by this catalog. [1] [5]

The deployment code provides a concrete controller path, including a Dynamixel client and a CUDA device selection. This is implementation evidence, not a verified USB connection or a minimum GPU-memory requirement. Check the selected policy loader, environment and hardware SDK together before attempting to transfer a trained policy. [3]

Match the object and interface

The upstream physical example uses a PLA cube measuring 7.5 cm on each side and weighing 281 g. That is a task condition, not evidence that arbitrary objects will reorient successfully. The README separates the direct Python SDK route from a ROS 2 Humble route and describes an example-node change for the latter. A setup for one route should not be treated as verification of the other. [1]

After SDK preparation, the ROS instructions direct users to comment out ros2_example.py in launch_leap.py before starting the task node. The README asks for port configuration, while the inspected direct Python controller tries /dev/ttyUSB0 through /dev/ttyUSB2. Its Deployment Notes request 5 V power and a micro-USB connection. These are unverified upstream instructions: reconcile the exact hand and SDK power requirements before connecting or energizing hardware. No electrical or port configuration was validated here. [1] [3]

The README declares an MIT license, but a standalone root license file was not identified in the inspected tree. Preserve that declaration and the separate rights of Isaac components, external SDKs and any selected checkpoint. No simulator, policy training, live transport or physical hand was operated here. [1] [4]

What we checked

Source research only. Selected implementation/model text, repository tree and license declarations were inspected. No upstream compilation, dependency installation, simulator, network/USB/device integration, data collection, benchmark or physical robot test was performed. Source files are linked rather than redistributed; external services and paid products are not endorsed.

Not yet established

  • Requirements: operating system, 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
c576e6a

Source files and licenses

Distinct robot-facing simulation contribution with actual implementation/model content and a bounded documented next step. Review covers source artifacts, not operational performance or a complete tested deployment.

LayerAvailability / license
softwareavailable · MIT declared in README; no standalone root license file identified
Declaration for the scoped source implementation/documentation; dependencies, external services, model/data assets and inherited file-specific terms require separate checks. No blanket redistribution or operational clearance. [1] [2] [3]
documentationavailable · MIT declared in README; no standalone root license file identified
Declaration for the scoped source implementation/documentation; dependencies, external services, model/data assets and inherited file-specific terms require separate checks. No blanket redistribution or operational clearance. [1] [2] [3]

This profile is scoped to source implementation/model descriptions and an initial inspection workflow; no physical manufacturing design, pretrained checkpoint or dataset is selected for redistribution or tested deployment.

References

Each source links to the exact revision we read. “Used for” shows where this page relies on it.

Correction guidance · Discussion is not enabled · Profile format 0.7.0-draft.5

Change history