search

Opentrons / opentrons

Software for writing protocols and running them on the Opentrons Flex and Opentrons OT-2
https://opentrons.com
Apache License 2.0
415 stars 177 forks source link

feat(api): allow some way of force ejecting tip even if the robot doesn't think it's attached #6243

Open theosanderson opened 4 years ago

theosanderson commented 4 years ago

Overview

Trying to run a protocol when a tip is already on a pipette (e.g. due to an abandoned previous protocol) leads to bad times, with collisions etc. Unfortunately one can't just add pipette.drop_tip to the top of a protocol because the drop_tip command asserts that the instrument has picked up a tip.

I've spent a while looking at hacking this and it's not very straightforward.

MarcelRobitaille commented 3 years ago

I've also tried to hack into thinking there was a tip attached so I could then call drop_tip. Why is this error even a thing? What does it hurt to eject without a tip attached?

mcous commented 3 years ago

Hello again @MarcelRobitaille! Fancy meeting you here.

This is pretty frustrating behavior. I think it's a symptom of tip tracking state living just a little bit lower in the stack than it probably should. The behavior comes from a place of the hardware controller needing to take tip length into account to understand how to move a given pipette. That fact snowballed into "the hardware controller keeps track of tip state" and then, without considering the larger implications, "the hardware controller should check if a tip is attached before trying to drop it".

This is a really good ticket to factor into some foundational protocol behavior work that is happening right now. In the meantime, though, I think we should see if we can remove that check in the hardware controller in a safe manner.

There really doesn't appear to be a good workaround for this; even if you go through private APIs it's messy (and brittle) enough for it to be a pretty bad idea

mattwelch commented 3 years ago

Thinking is that fixing #5248 will make this moot in practice, though a fundamental fix would still be nice.

MarcelRobitaille commented 3 years ago

@mattwelch why does that make it moot?

mcous commented 3 years ago

@MarcelRobitaille sorry, this looks like a potential typo or miscommunication. Feel free to disregard; unentangling the tip drop procedure from tip attached state is still on our radar