PID controllers are incredibly powerful tools for managing dynamic systems - from keeping a heater at the perfect temperature, to balancing a robot on two wheels, to stabilising a drone in flight.
In my optimal egg machine for example, I use a PID loop to continuously compare the target temperature to the actual temperature, adjusting the heater’s power in real-time to maintain precise control over the heating process.
But getting this to work was a pain - while I was eventually able to get a PID autotuning library to work, it still involved a lot of manual variable tweaking which felt pretty self-defeating.
Yesterday, while struggling yet again to tune a different heater, I decided I’d had enough. I wanted something that completely abstracts away the frustration of fiddling with magic numbers, so I could focus on actually building my project.
For that reason, I quickly hacked together Vibe PID - an intelligent PID tuning assistant for the 21st Century!
Vibe PID
TL;DW:
You don’t need to sign up for an account
Just enter your OpenAI API key in the settings page. Your API key stays on your local machine - I can’t access your private key. The code is open source so you can check for yourself!
You can iterate on PID variables using smart reasoning models like o4-mini.
The website helps organise all of the context and generates visualisations so the LLM can make intelligent parameter adjustment suggestions.
Next Steps
I hope this is useful. I am new to control engineering so this implementation may be rough around the edges. The code is open source, so if there are any improvements you’d like to make feel free just make a PR or add an issue! You can also get in touch on Twitter.