What is the Zeroth Bot platform?

Zeroth Bot (Z-Bot) is an affordable, open-source 12-DOF humanoid robot platform designed for researchers and hobbyists to explore Reinforcement Learning and sim-to-real transfer.

How does simulation work with Z-Bot?

Z-Bot uses a MuJoCo-based training pipeline (ksim) to train Reinforcement Learning policies in a physics-accurate simulation before deploying them to the physical hardware.

What hardware is required to build a Zeroth Bot?

The BOM includes 12 standard hobby servos, an ESP32 or Teensy microcontroller, an IMU (like the MPU6050), a 2S-3S LiPo battery, and a set of 3D-printed parts.

What is the estimated cost of the Zeroth Bot?

The Bill of Materials (BOM) for Zeroth Bot is approximately $500, making it a highly accessible entry point for humanoid robotics experimentation.

Can I control Zeroth Bot through a web browser?

Yes, Zeroth Bot features a web-based frontend that provides real-time telemetry visualization, manual joint control, and policy deployment interfaces.

Zeroth Bot Robotics Platform

Name: Zeroth Bot Robotics Platform
Author: plurigrid

byplurigrid

•

Data Science & ML

Facilitates the development, simulation, and control of 3D-printed humanoid robots for reinforcement learning research.

Zeroth Bot (Z-Bot) is an accessible, open-source humanoid robotics platform specifically engineered to bridge the gap between simulation and reality. As a cost-effective 12-DOF robot, it provides researchers and developers with a complete ecosystem for training Reinforcement Learning (RL) policies using MuJoCo, deploying firmware via ESP32 or Teensy, and managing hardware through a dedicated web frontend. It is an ideal entry point for rapid prototyping and educational robotics, offering a streamlined pipeline from 3D printing components to executing complex walking behaviors in the real world.

Key Features

0112-Degree-of-Freedom (DOF) humanoid design for complex movement research

02Low-cost 3D-printable frame optimized for PLA/PETG and standard hobby servos

032 GitHub stars

04Web-based control interface for real-time telemetry and manual joint calibration

05Full MuJoCo integration for seamless sim-to-real Reinforcement Learning training

06Standardized hardware BOM for an affordable (~$500) entry into humanoid robotics

Use Cases

01Training and validating walking policies in simulation for physical humanoid deployment

02Rapid prototyping of humanoid behaviors and hardware-in-the-loop experimentation

03Academic research and education in robotics, kinematics, and control systems

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add plurigrid/asi zeroth-bot

For use in Claude.ai and ChatGPT

Download Skill