The unified API layer for robotics. Connect any robot, any brand, with one SDK.
The Problem
Every robot manufacturer ships a different SDK, a different protocol, and a different data format.
A Boston Dynamics Spot speaks nothing like a Universal Robots UR5.
A Figure humanoid has nothing in common with an Agility Robotics Digit.
Every team building on top of robots rewrites the same integration layer from scratch.
This is a multi-billion dollar tax on the robotics industry.
The Solution
RoboAPI is the Stripe for Robotics — a single unified API that abstracts every robot into one clean developer experience. One SDK. One API key. Every robot.
import roboapi # Connect to ANY robot — simulated, ROS2, or real hardware robot = roboapi.connect(robot_id="spot-01", brand="simulated") # Universal commands — same code regardless of robot brand robot.move(x=1.0, y=0.0, speed=0.5) # Live telemetry — normalised across all hardware telemetry = robot.sense() print(f"Position: {telemetry.position}") print(f"Battery: {telemetry.battery_pct}%") # Works identically on a Boston Dynamics robot, a ROS2 robot, or a simulation
Quickstart — 5 minutes to your first robot
1. Clone and install
git clone https://github.com/YOUR_USERNAME/roboapi.git cd roboapi conda create -n roboapi python=3.11 -y conda activate roboapi pip install fastapi "uvicorn[standard]" pydantic websockets httpx python-dotenv roslibpy
2. Start the server
uvicorn api.main:app --reload
🤖 RoboAPI v0.1.0 starting up...
INFO: Uvicorn running on http://127.0.0.1:8000
3. Connect a simulated robot
curl -X POST http://localhost:8000/v1/robots/connect \ -H "Content-Type: application/json" \ -d '{"robot_id": "my-robot", "brand": "simulated"}'
{
"robot_id": "my-robot",
"brand": "simulated",
"status": "connected",
"connected_at": 1714000000.0
}4. Move it
curl -X POST http://localhost:8000/v1/robots/my-robot/move \ -H "Content-Type: application/json" \ -d '{"x": 1.0, "y": 0.0, "speed": 0.5}'
5. Read live telemetry
curl http://localhost:8000/v1/robots/my-robot/sense
{
"robot_id": "my-robot",
"battery_pct": 99.8,
"position": {"x": 0.037, "y": 0.001, "z": 0.0},
"velocity": {"vx": 0.749, "vy": 0.002, "vz": 0.0},
"status": "busy"
}That's it. No ROS knowledge required. No brand-specific SDKs. No protocol translation.
Browse the full interactive API docs at http://localhost:8000/docs
Connecting a Real ROS2 Robot
RoboAPI connects to any ROS2 robot via rosbridge.
# Terminal 1 — start your ROS2 robot (or turtlesim for testing) ros2 run turtlesim turtlesim_node # Terminal 2 — start rosbridge ros2 launch rosbridge_server rosbridge_websocket_launch.xml # Terminal 3 — start RoboAPI uvicorn api.main:app --reload
# Connect to the ROS2 robot through RoboAPI curl -X POST http://localhost:8000/v1/robots/connect \ -H "Content-Type: application/json" \ -d '{ "robot_id": "turtle1", "brand": "ros", "host": "localhost", "port": 9090 }' # Move it — RoboAPI translates to ROS Twist messages automatically curl -X POST http://localhost:8000/v1/robots/turtle1/move \ -H "Content-Type: application/json" \ -d '{"x": 1.0, "y": 0.8, "speed": 0.6, "duration_ms": 6500}'
The turtle draws a full circle — driven entirely through the unified RoboAPI layer. 🐢
API Reference
Robots
| Method | Endpoint | Description |
|---|---|---|
POST |
/v1/robots/connect |
Connect a robot and register it |
POST |
/v1/robots/{id}/move |
Send a normalised move command |
POST |
/v1/robots/{id}/stop |
Graceful or emergency stop |
GET |
/v1/robots/{id}/sense |
Poll current sensor snapshot |
GET |
/v1/robots/{id}/status |
Connection info and health |
DELETE |
/v1/robots/{id}/disconnect |
Disconnect and deregister |
GET |
/v1/robots |
List all connected robots |
Fleet
| Method | Endpoint | Description |
|---|---|---|
GET |
/v1/fleet |
List all robots in fleet |
POST |
/v1/fleet/stop-all |
Emergency stop every robot |
POST |
/v1/fleet/broadcast/move |
Send move to all robots at once |
Telemetry
| Method | Endpoint | Description |
|---|---|---|
WS |
/v1/telemetry/{id}/stream |
Real-time stream at configurable Hz |
# WebSocket telemetry stream — live position updates at 5Hz wscat -c "ws://localhost:8000/v1/telemetry/my-robot/stream?hz=5"
Supported Robots
| Brand | Status | Adapter |
|---|---|---|
| Simulated | ✅ Ready | Built-in physics sim, battery, sensor noise |
| ROS / ROS2 | ✅ Ready | Any ROS2 robot via rosbridge WebSocket |
| Boston Dynamics | 🔶 Planned | Phase 2 |
| Universal Robots | 🔶 Planned | Phase 2 |
| Figure / Humanoids | 🔶 Planned | Phase 2 |
| Custom hardware | 🔶 Planned | Open adapter spec — implement BaseAdapter |
Adding a new robot brand
RoboAPI uses a pluggable adapter pattern. To add any robot:
# adapters/your_robot_adapter.py from adapters.base import BaseAdapter class YourRobotAdapter(BaseAdapter): async def connect(self) -> bool: ... async def disconnect(self) -> None: ... async def move(self, cmd: MoveCommand) -> CommandResponse: ... async def stop(self, cmd: StopCommand) -> CommandResponse: ... async def get_telemetry(self) -> TelemetrySnapshot: ... async def is_alive(self) -> bool: ...
Register it in adapters/__init__.py and it's immediately available through the full API.
Architecture
┌─────────────────────────────────────────────────┐
│ Your Application │
│ (curl / Python SDK / Dashboard) │
└─────────────────┬───────────────────────────────┘
│ REST + WebSocket
┌─────────────────▼───────────────────────────────┐
│ RoboAPI Core │
│ ┌──────────┐ ┌──────────┐ ┌───────────────┐ │
│ │ Robots │ │ Fleet │ │ Telemetry │ │
│ │ API │ │ API │ │ WebSocket │ │
│ └──────────┘ └──────────┘ └───────────────┘ │
│ ┌─────────────────────────────────────────────┐ │
│ │ Adapter Factory │ │
│ └─────────────────────────────────────────────┘ │
└────┬──────────────┬──────────────┬───────────────┘
│ │ │
┌────▼────┐ ┌──────▼──────┐ ┌───▼──────────────┐
│ Sim │ │ ROS2 │ │ Future: BD / UR │
│ Adapter │ │ Adapter │ │ Adapters │
└─────────┘ └─────────────┘ └──────────────────┘
│ │
┌────▼────┐ ┌──────▼──────┐
│ Built- │ │ rosbridge │
│ in │ │ port 9090 │
│ Sim │ └──────▼──────┘
└─────────┘ │
┌────▼────────────────┐
│ Any ROS2 Robot │
│ Turtlesim · Spot · │
│ Custom hardware │
└─────────────────────┘
Project Structure
roboapi/
├── api/
│ ├── main.py # FastAPI app entry point
│ └── v1/
│ ├── robots.py # Core robot endpoints
│ ├── fleet.py # Fleet management
│ └── telemetry.py # WebSocket streaming
├── adapters/
│ ├── base.py # Abstract adapter interface
│ ├── sim_adapter.py # Simulated robot
│ └── ros_adapter.py # ROS2 via rosbridge
├── core/
│ ├── command.py # Pydantic models
│ └── robot_registry.py # Robot state management
├── sdk/ # Python SDK (coming soon)
├── dashboard/ # React dashboard (coming soon)
└── tests/
└── test_api.py # 14 passing tests
Roadmap
Phase 1 — Foundation ✅ (current)
- Unified REST API — connect, move, stop, sense
- Simulated robot adapter
- ROS2 adapter via rosbridge
- Real-time WebSocket telemetry
- Fleet management — stop-all, broadcast
- 14-test pytest suite
- Auto-generated API docs
Phase 2 — Platform (next)
- API key authentication
- Python SDK —
pip install roboapi - React developer dashboard
- Boston Dynamics adapter
- OTA firmware update pipeline
- Redis-backed fleet registry
Phase 3 — Intelligence
- AI bridge — connect any LLM to any robot
- Natural language commands
- Vision model integration
- Data flywheel — aggregate telemetry model
Phase 4 — Scale
- Hosted cloud platform
- Multi-tenant fleet management
- Compliance & audit logging
- Enterprise SLAs
Development
# Run tests pytest tests/ -v # Run with auto-reload uvicorn api.main:app --reload # Code formatting black . ruff check .
Contributing
RoboAPI is in early development and we welcome contributions — especially:
- Robot adapters — if you have access to hardware we haven't supported yet
- Bug reports — open an issue with your robot model and error
- Feature requests — what would make this useful for your project
See CONTRIBUTING.md for guidelines.
Why RoboAPI?
The robotics industry is at its Stripe moment.
Before Stripe, every company built custom payment integrations. Before Twilio, everyone wrote their own SMS stack. The pattern is always the same: fragmented, complex, infrastructure problem → one abstraction layer wins → everything builds on top.
Robotics is there right now. Dozens of manufacturers, dozens of protocols, thousands of teams rebuilding the same middleware. RoboAPI is the abstraction layer.
License
MIT — see LICENSE
Status
🚧 Early development — API may change. Not yet recommended for production.
⭐ Star this repo to follow progress.
💬 Open an issue to request a robot adapter or report a bug.























