Node Types

The four types of nodes in the OctoMY™ network, each serving a distinct role.

Did You Know?

The term "Agent" was chosen instead of "robot" because OctoMY™ can control anything with actuators and sensors - not just physical robots. A smart home device, an animatronic prop, or even a virtual character could all be Agents.

Agent

The Agent represents a robot or autonomous device. It is called an "Agent" rather than "robot" because the term is more general and doesn't impose assumptions about physical form.

Characteristics

• Runs on the robot hardware itself
• Controls actuators (motors, servos, etc.)
• Reads sensors (cameras, distance, orientation)
• Executes Plans for autonomous behavior
• Can run a local LLM for on-device intelligence

Typical hardware

• Android phone attached to a robot chassis
• Raspberry Pi with motor controllers
• Arduino with communication shield
• Custom embedded system

Examples

• RC car with Arduino + Android combo
• Hexapod walking robot
• Quadcopter drone
• Virtual robot in a simulation

Remote

The Remote represents the user interface for controlling Agents. It allows both human operators and automated systems to provide input.

Characteristics

• Thin client focused on user interface
• Displays real-time status from Agents
• Sends commands to paired Agents
• Adapts UI dynamically based on Agent capabilities
• Can control multiple Agents simultaneously

Input modes

Remotes support various levels of control:

• Low-level: Direct actuator manipulation in real-time
• Mid-level: Preset movements and gestures
• High-level: Target points on maps, natural language commands

Typical hardware

• Smartphone or tablet
• Desktop computer
• Dedicated control station

Hub

The Hub is a coordination node designed for stationary, high-performance hardware. It serves as the bridge between the local OctoMY™ network and external services.

Characteristics

• Runs on server or workstation hardware
• Never runs on mobile devices
• Gateway to internet services
• Coordinates multi-Agent operations
• Provides "LLM as a service" to Agents

Responsibilities

• Data aggregation: Collects data from all connected Agents
• Heavy processing: Runs compute-intensive tasks (AI inference, path planning)
• External integration: Connects to cloud APIs, external LLMs
• Fleet coordination: Manages collaboration between multiple Agents
• Persistent storage: Long-term data archival

Key principle

Agents and Remotes never contact the internet directly. All external communication goes through a Hub, creating a single point of security control.

Zoo

The Zoo is OctoMY™'s discovery and connectivity service. It helps nodes find each other across network boundaries.

Characteristics

• Runs as a cloud service
• Minimal functionality for security
• No access to user data
• Facilitates NAT traversal

Services

• UDP hole-punching: Helps nodes connect through firewalls
• Discovery coordination: Helps nodes find each other by GPS proximity
• Connectivity relay: Last resort when direct connection fails

Security design

The Zoo is designed to know as little as possible:

• No access to message contents
• No persistent storage of user data
• Minimal attack surface

Node comparison

Network topology

A minimal OctoMY™ setup requires only an Agent and a Remote. Additional nodes add capabilities:

Minimal setup

With Hub

Full network