Zeroth Bot Skill Skill | Agent Skills

Zeroth Bot Skill

Trit: -1 (MINUS - specification/verification) Color: #8CC136 (Lime Green) URI: skill://zeroth-bot#8CC136

Overview

Zeroth Bot (Z-Bot) is a 3D-printed open-source humanoid robot platform designed for sim-to-real research and RL experimentation. An affordable entry point for humanoid robotics.

Specifications

┌────────────────────────────────────────────────────────────────┐
│                      ZEROTH BOT (Z-BOT)                        │
├────────────────────────────────────────────────────────────────┤
│                                                                 │
│  Height: ~40cm                                                 │
│  Weight: ~2kg                                                  │
│  DOF: 12 joints                                                │
│                                                                 │
│  Frame: 3D printed (PLA/PETG)                                  │
│  Actuators: Servo motors                                        │
│  Cost: ~$500 BOM                                               │
│                                                                 │
│  Ideal for:                                                     │
│  ├── Learning sim-to-real transfer                             │
│  ├── Testing RL policies at low cost                           │
│  ├── Educational robotics                                      │
│  └── Rapid prototyping                                         │
│                                                                 │
└────────────────────────────────────────────────────────────────┘

Hardware BOM

| Component | Quantity | Notes | |-----------|----------|-------| | Servo motors | 12 | Standard hobby servos | | 3D printed parts | Full set | STL files provided | | MCU | 1 | ESP32 or Teensy | | IMU | 1 | MPU6050 or similar | | Power | 1 | 2S-3S LiPo |

Training Pipeline

from ksim.robots.zbot import ZBotConfig
from ksim import PPOTask

class ZBotWalking(PPOTask):
    robot = ZBotConfig(
        model_path="zbot.mjcf",
        servo_config={
            "kp": 50.0,
            "kd": 5.0,
            "torque_limit": 5.0,  # Smaller than K-Bot
        }
    )
    
    # Faster training due to simpler robot
    training_config = {
        "num_envs": 2048,
        "learning_rate": 5e-4,
    }

GF(3) Triads

zeroth-bot (-1) ⊗ kos-firmware (+1) ⊗ mujoco-scenes (0) = 0 ✓

Related Skills

kbot-humanoid (-1): Larger flagship humanoid
ksim-rl (-1): RL training library
kos-firmware (+1): Firmware (kos-zbot variant)
urdf2mjcf (-1): Model conversion

Web Frontend

Z-Bot has a web-based control interface:

Real-time telemetry visualization
Manual joint control
Policy deployment interface

// From zbot-web-frontend
import { ZBotController } from '@kscale/zbot-web';

const controller = new ZBotController({
  endpoint: 'ws://zbot.local:8080',
});

await controller.connect();
await controller.setJointPositions({
  hip_pitch_l: 0.5,
  knee_l: -0.3,
});

References

@misc{zerothbot2024,
  title={Zeroth Bot: 3D-Printed Open-Source Humanoid},
  author={K-Scale Labs},
  year={2024},
  url={https://github.com/kscalelabs/zeroth-bot}
}

Agent Skills: Zeroth Bot Skill

Install this agent skill to your local

Skill Files