Open Source RoboticsRobots, software & learning

Catalog

Robots, end effectors, software and robotics models: compare setup requirements, configuration costs, checkpoints and source licenses.

Explore three documented starting workflows

Choose by task

Three ways to start

Learning routes →

Selected for their documented workflows. Source researched; we have not tested these setups.

Teleoperate an arm

SO-101 →

Compare follower-only and leader–follower configurations, then follow the pinned build and calibration route.

Printed frame, compatible host computer and calibration on the actual arms.

Start in simulation

Webots →

Follow the documented introductory tutorial to build a virtual world and run a controller.

Check the supported operating system and installation requirements before starting.

Explore robot learning

LeRobot →

Begin with an existing dataset to inspect observations and actions before choosing physical hardware.

Python and workflow-specific dependencies; training needs a separate compute budget.

440 entries · Newest published first

Source researched · Not hands-on tested

Vision-language-action (VLA) models · Model

DexVLA

Vision-Language-Action architecture pairing an off-the-shelf Qwen2-VL-2B-Instruct backbone with a plug-in ScaleDP diffusion transformer action expert (~1B ScaleDP-H or ~400M/410M ScaleDP-L), releasing Stage-1 pretrained ScaleDP weights (CC-BY-NC-4.0) and three-stage training code under Attribution-NonCommercial terms.

~3B · Code: Creative Commons Attribution-NonCommercial summary terms · Weights: CC-BY-NC-4.0

ManipulationRobot learning
2 checkpoints

Locomotion policies · Model

ExBody (Expressive Whole-Body Control for Humanoid Robots)

Apache-2.0 Isaac Gym RL whole-body humanoid controller for the 19-DoF Unitree H1 that tracks retargeted CMU Mocap upper-body motions while relaxing lower-body constraints to robust root velocity tracking; upstream ships an in-repo pretrained H1 checkpoint (060-40/model_49500.pt), TorchScript export, and ASE-based motion retargeting scripts.

Compact RL MLP (~11.8 MB PyT · Code: Apache-2.0 · Weights: Unresolved

LocomotionRobot learning
1 checkpoint

Locomotion policies · Model

H2O & OmniH2O (Human-to-Humanoid Whole-Body Control)

CC-BY-NC-4.0 (non-commercial) Isaac Gym sim-to-real RL and DAgger distillation codebase for Unitree H1 whole-body humanoid teleoperation and autonomy (H2O and OmniH2O); upstream releases motion retargeting scripts, filtered AMASS motion data, and hardware wrappers, but not pretrained humanoid policy checkpoints.

Config-dependent compact MLP · Code: CC-BY-NC-4.0 · Weights: Unresolved

LocomotionTeleoperationRobot learning
1 checkpoint

Visuomotor policies · Model

HumanPlus (HST & HIT Humanoid Shadowing and Imitation)

Full-stack Stanford humanoid learning repository combining the Humanoid Shadowing Transformer (HST, Isaac Gym PPO low-level whole-body controller) and Humanoid Imitation Transformer (HIT, ACT/DETR-derived stereo RGB visuomotor behavior cloning policy) plus Unitree H1 hardware code and Google Drive demonstrations; no root license or pretrained checkpoints are published.

Not published as a single sc · Code: Unresolved · Weights: Unresolved

ManipulationLocomotionTeleoperation
1 checkpoint

Visuomotor policies · Model

Improved 3D Diffusion Policy (iDP3)

MIT-licensed egocentric 3D visuomotor diffusion policy for humanoid manipulation that operates on unsegmented, uncalibrated 4096-point LiDAR/depth point clouds via a multi-stage PointNet encoder and 1D conditional U-Net DDIM action head; upstream releases training/deployment code, datasets, and 3-task Fourier GR1 checkpoints on Google Drive.

Task-dependent (not publishe · Code: MIT · Weights: Unresolved

ManipulationRobot learning
1 checkpoint

Visuomotor policies · Model

VQ-BeT (Vector-Quantized Behavior Transformer)

MIT-licensed multimodal behavior generation policy that tokenizes continuous actions with a hierarchical Residual VQ-VAE and predicts discrete latent action codes plus continuous offsets with a causal GPT transformer; upstream releases training/evaluation code, benchmark datasets, and pretrained goal-conditioned Kitchen weights on Google Drive.

Task-dependent (not publishe · Code: MIT · Weights: Unresolved

ManipulationRobot learning
1 checkpoint

Visuomotor policies · Model

3D Diffusion Policy (DP3)

MIT-licensed 3D visuomotor diffusion policy architecture combining a compact PointNet-style sparse point-cloud encoder with a 1D conditional U-Net DDIM action head for data-efficient dexterous and arm manipulation; upstream releases training/evaluation code, RL demonstration experts, and real-robot datasets, but not pretrained DP3 policy weights.

Task-dependent (not pinned a · Code: MIT · Weights: Unresolved

ManipulationRobot learning
1 checkpoint
Other known designs · listed, not profiled

Listed, not profiled

We know about these designs but do not profile them. Each note says why. Follow the links for the official project and its license.

Platform tags identify documented interfaces or applications; they do not establish tested fit or certification. Methodology · About the spec