Give Embodied AI a Real Soul. PhanthyMotus is a next-generation, open-source framework and platform for Embodied AI Agents. Built upon a robust ROS2 foundation, it seamlessly bridges diverse sensor inputs with advanced robot execution. By enabling flexible integration of World Models, LLMs, and VLMs, PhanthyMotus transforms traditional hardware into soulful, intelligent assistants capable of perceiving, thinking, and acting independently in the real world.
Install and run with a single command:
curl -fsSL https://motus.phanthy.com/install.sh | sudo bashOr specify a version:
curl -fsSL https://motus.phanthy.com/install.sh | sudo bash -s <tag>The install script will automatically install Docker (if needed), pull the latest Agent Core image, and start the service.
Open http://<device-ip>:15678 to access the Web Dashboard.
Browse available versions and images at the Resource Center.
Deploy hardware drivers from phanthymotus-driver. Drivers automatically register with Agent Core on startup — no manual configuration needed.
See CONTRIBUTING.md for building and running from source code.
- Visual Orchestration — Drag-and-drop web dashboard for connecting devices, sensors, and AI models on a canvas
- MCP Data Bus — Unified Model Context Protocol interface for all hardware devices
- Event-Driven Agent Loop — LLM-powered reasoning with multi-turn tool calling, driven by real-time sensor events
- ROS2 Integration — Native DDS bridge for seamless ROS2 topic relay and monitoring
- Pluggable Perception — Modular ASR/TTS stack with local inference support (Jetson)
- Web Dashboard — Real-time device monitoring, agent activity stream, and configuration — all from the browser
Hardware drivers are maintained in a separate repository: phanthymotus-driver.
The dashboard at http://<device-ip>:15678 provides:
Add sensors and actuators you need onto the canvas, connect them to the core Agent Loop, and the framework handles data flow and execution automatically. Build your embodied AI agent like stacking building blocks.
Live sensor data visualization — audio waveforms, battery status, 3D skeleton/point cloud, and more.
Define the agent's identity, system prompt, and long-term memory directly from the UI.
Browse past agent sessions with full event traces and tool call results.
A community-driven Skill Marketplace where users share and discover skills. Browse and install skills contributed by others, or teach your robot new capabilities using natural language — no coding required.
Deploy and manage Agent Core and hardware driver containers from the dashboard.
| Service | Port |
|---|---|
| Agent Core | 15678 |
| Perception MCP | 15720 |
| Perception WebSocket | 15721 |
Hardware driver ports are documented in phanthymotus-driver.
The platform can optionally connect to a Resource Center for:
- Browsing and deploying pre-built driver/perception images
- Managing skills and extensions
- OTA updates
Configure via the RESOURCE_CENTER_URL environment variable.
See CONTRIBUTING.md for development setup, architecture details, and guidelines.