strands-labs

strands-labs / robots-sim

Public

Simulated environments for robot agent evaluation and reinforcement learning.

strandsagents.com agentic agentic-ai ai genai machine-learning
20
4
100% credibility
Found Feb 24, 2026 at 14 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

A simulation toolkit letting people control virtual robots with everyday language through smart AI helpers, capturing videos of tasks for easy review.

How It Works

1
🔍 Discover Safe Robot Playground

You find a fun way to test robot ideas in a virtual world without needing real machines.

2
🛠️ Set Up Your Virtual Workshop

You prepare the pretend robot world on your computer so everything is ready to play.

3
🧠 Connect the Smart Robot Brain

You link a clever thinking helper that understands pictures and words to guide the robot.

4
🗣️ Tell Robot What to Do

You speak or type simple instructions like 'pick up the red block' and watch it happen step by step.

5
📹 See Videos of Actions

You get fun side-by-side videos showing exactly what the robot did from different views.

🎉 Master Robot Skills Safely

You quickly test and improve robot tricks in simulation, ready to try on real hardware anytime.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 14 to 20 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 robots-sim?

Robots-sim is a Python framework for creating simulated environments to evaluate robot agents and reinforcement learning policies, especially in Libero benchmarks. It lets you control simulated robots via natural language instructions using vision-language-action models like GR00T, with automatic video rollouts saved to `./rollouts/`. Developers get rapid prototyping without hardware, via simple CLI examples like `python examples/libero_example.py` for full episodes or stepped mode for iterative control.

Why is it gaining traction?

It stands out with dual modes—full autonomous runs or stepped execution returning camera images and state for agent feedback—plus batch scripts for success-rate logging. Policy abstraction swaps VLAs easily (GR00T via ZMQ out-of-box), and video recording with side-by-side views beats basic robots simulation tools. For recoworld building simulated environments for agentic systems, the Strands agent integration hooks LLM-driven planning without custom glue code.

Who should use this?

Robot RL researchers benchmarking VLAs on Libero's 90-task suite, agent engineers prototyping hierarchical control (System 1 execution + System 2 planning), or teams at robots sim companies testing policies pre-hardware. Ideal if you're tired of manual sim setup for natural language robot tasks.

Verdict

Try it for Libero+GR00T workflows—docs, examples, and tests are solid despite 11 stars and 1.0% credibility signaling early maturity. Production use needs more adoption, but it's a practical simulator for sim-first dev.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.