Software Developer Kit

The Dropbear SDK bundles CLI tools, project templates, simulation launchers, and debugging utilities into a single developer toolkit. It's the fastest way to go from zero to a running robot — real or simulated.

Installation

# Install the SDK (includes CLI + Python API + simulation tools)
pip install dropbear-sdk

# Verify installation
dropbear --version

CLI Reference

The dropbear CLI provides commands for project scaffolding, simulation, deployment, and diagnostics.

Command	Description
dropbear init	Scaffold a new Dropbear project with ROS 2 workspace structure
dropbear sim start	Launch the MuJoCo or Gazebo simulation environment
dropbear sim teleop	Open teleoperation control (gamepad or web UI)
dropbear train	Launch an RL training run in Isaac Lab
dropbear deploy	Deploy a trained policy to hardware or simulation
dropbear skill publish	Package and publish a skill to the marketplace
dropbear doctor	Check system dependencies and configuration

Project Templates

Quickly bootstrap new projects with built-in templates:

# Create a new skill project
dropbear init --template skill my-walking-skill

# Create a new perception pipeline
dropbear init --template perception my-vision-pipeline

# Create a full ROS 2 package
dropbear init --template ros2-package my-custom-controller

Simulation Tools

The SDK includes pre-configured simulation environments:

MuJoCo — Fast physics for RL training and rapid iteration. Runs headless or with a viewer.
Gazebo — Full ROS 2 integration for system-level testing with realistic sensors.
Isaac Lab — GPU-accelerated parallel simulation for large-scale RL training.

# Launch MuJoCo with the Dropbear MJCF model
dropbear sim start --engine mujoco

# Launch Gazebo with ROS 2 bridge
dropbear sim start --engine gazebo --ros2

# Headless training mode (no rendering)
dropbear sim start --engine isaac --headless --num-envs 4096

Debugging & Diagnostics

# Run system diagnostics
dropbear doctor

# Stream joint data to terminal
dropbear monitor joints

# Record a rosbag for replay
dropbear record --topics /joint_states /imu /camera/image_raw

# Replay a recorded session in simulation
dropbear replay session_2026_07_03.bag

🚧 SDK Status

The SDK is under active development. Core simulation and project scaffolding commands are functional. Training and deployment commands are being refined. Star the GitHub repo to follow progress.