Python SDK (preview)#

Planned - not yet available

The Python Swarm SDK described on this page is a design preview, not shipped code. None of the snippets below run today - they show the API we intend to build. The imports (from dotbot import Swarm) and every method (Swarm.connect, Swarm.run, bot.move_to, …) are aspirational.

To script the swarm today, use the CLI: start a controller with dotbot run controller, then drive bots over its REST / WebSocket surface (or the MQTT bridge).

What it will be#

The SDK will be a thin Python wrapper over a running controller’s REST/WS surface, so you write swarm logic in Python instead of hand-rolling HTTP and asyncio. You start a controller once (dotbot run controller), then a script connects to it and commands bots. The same script targets real hardware, the simulator, or a remote testbed - the backend is chosen at run time, not in the code.

Intended API#

All three snippets are aspirational - they will not run until the SDK ships.

Connect and drive one bot - connect to a local controller, grab a bot, set its color, move it:

from dotbot import Swarm

async with Swarm.connect() as swarm:        # defaults to http://localhost:8000
    bot = next(iter(swarm))
    bot.set_color(red=255)
    await bot.move_to(500, 500)

Run an algorithm - Swarm.run() handles argv parsing and asyncio.run(), so a student writes only the algorithm body:

from dotbot import Swarm

async def algorithm(swarm):
    for bot in swarm:
        bot.set_color(red=255)

if __name__ == "__main__":
    Swarm.run(algorithm)

Switch backends without editing code - the same script runs against the local default, the simulator, or a remote class testbed, picked by a CLI flag:

python my_assignment.py                                   # local controller
python my_assignment.py --sim                             # simulator
python my_assignment.py --swarm-url http://classroom:8000 # shared testbed

Until then#

On this page
Show Source