qiaoguanren

qiaoguanren / Signbot

Public

[ICRA 2026] SignBot: Learning Human-to-Humanoid Sign Language Interaction

19
1
100% credibility
Found Mar 26, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

SignBot trains humanoid robots to perform and interact using sign language motions through motion retargeting and reinforcement learning.

How It Works

1
🔍 Discover SignBot

You stumble upon SignBot while watching cool robot videos and reading about AI that understands sign language.

2
📱 Watch the Demo

Click the video link to see a humanoid robot smoothly signing words like a pro, sparking your excitement to try it.

3
🛠️ Set Up Your Workspace

Follow simple guides to prepare your computer, like installing everyday tools so everything runs smoothly.

4
📥 Gather Sign Language Motions

Download real sign language videos and convert them into robot-friendly movements with easy scripts.

5
🤖 Train Your Signing Robot

Launch training where the robot learns to mimic signs perfectly, watching it improve over time like magic.

6
▶️ Test and Play

Run play mode to see your robot perform signs fluidly, tweaking speeds and motions for fun.

7
Add Conversation
🗣️
Chat with Robot

Link an AI helper so the robot responds to your words in sign language.

Create New Signs

Use tools to invent fresh signs for custom interactions.

🎉 Robot Signs Fluently

Your humanoid now chats in sign language naturally, opening doors to amazing human-robot talks!

Sign up to see the full architecture

6 more

Sign Up Free

Star Growth

See how this repo grew from 19 to 19 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 Signbot?

SignBot trains humanoid robots like Unitree H1 to mimic human sign language motions in simulation, bridging human-to-humanoid interaction via RL policies in Isaac Gym. Python-based, it handles dataset prep from SMPLX sign language data (ASL/CSL), retargeting to robot kinematics, policy training with adversarial motion priors, and playback for signbot gif maker demos. Users run CLI commands like `python train.py` for custom policies or `python play.py` with pretrained models, plus FastAPI endpoints for LLM-powered sign translation/response/generation.

Why is it gaining traction?

Tied to an ICRA 2026 accepted paper on signbot learning human-to-humanoid sign language interaction, it stands out in icra papers github repos by offering reusable motion imitation for expressive robots—beyond basic locomotion. Devs dig the quick sim-to-real pipeline (tested on wheeled bots), wandb logging, and web viz for headless play, making icra 2026 vienna demos feasible without from-scratch RL setup.

Who should use this?

Robotics PhDs prepping ICRA 2026 submissions or workshops, humanoid RL teams retargeting mocap for dexterous tasks, accessibility researchers embodying sign language (ASL hybrids) on sim hardware like H1. Ideal for icra+github explorers prototyping signet online translators.

Verdict

Promising research tool at 1.0% credibility (19 stars, solid README but sparse tests/docs)—grab for ICRA 2026 program tinkering, but expect setup tweaks; fork and contribute for maturity.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.