InternRobotics

InternRobotics / SIM1

Public

Official implementation of "SIM1: Physics-Aligned Simulator as Zero-Shot Data Scaler in Deformable Worlds "

internrobotics.github.iosim1.github.io deformable-manipulation lerobot robotics simulation vision-language-action-model
83
6
100% credibility
Found Apr 13, 2026 at 89 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

SIM1 is an open-source physics simulator for generating synthetic training data of dual-arm robots manipulating deformable cloth like shirts.

How It Works

1
🧑‍🔬 Discover SIM1

You find a free simulator that lets robots handle squishy cloth like shirts with two arms, perfect for training AI without real hardware.

2
🛠️ Set up your simulator

Download the tools and assets, then launch it with a simple command – everything installs automatically like a game.

3
⌨️ Drive the robot arms

Use your keyboard to teleoperate the dual arms, folding and lifting shirts just like watching a clever puppet show.

4
🤖 Generate training examples

Create hundreds of cloth manipulation videos automatically, scaling up demos into diverse practice data for AI learning.

5
🎥 Review and polish

Replay motions, filter out glitches, and render photorealistic videos to make everything look like real-world footage.

📦 Export robot datasets

Get ready-to-train datasets that teach AI to manipulate cloth, saving time and hardware for robotics research.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 89 to 83 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is SIM1?

SIM1 delivers a physics simulator for dual-arm cloth manipulation, scaling synthetic data zero-shot for deformable environments. Developers get interactive teleoperation via keyboard or WebSocket, diffusion-based trajectory generation, replay with optional randomization, quality filtering, photoreal rendering, and LeRobot dataset export—all in Python using Newton and Warp. Run bash run_pipeline.sh --num 100 for instant trajectories, no manual stitching needed.

Why is it gaining traction?

One-command pipelines crush manual data workflows: generate, smooth with Kalman, replay USD/NPZ, filter unreachable joints or poor cloth alignment. Official GitHub repository bundles assets via download_assets.sh from Hugging Face, including pretrained diffusion models. Stands out for deformable sim fidelity without real-robot hours, plus remote teleop for headless setups.

Who should use this?

Robotics engineers prototyping dual-arm policies on cloth folding or lifting. RL teams short on real data, seeking sim-to-real via LeRobot exports. Researchers tweaking deformables, like sim1 einrichten for quick local pipelines.

Verdict

Grab it for cloth sim experiments—mature pipeline despite 83 stars and 1.0% credibility score. Docs shine in README quickstarts, but expect tweaks for production; monitor official GitHub releases page for Newton upgrades.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.