starting a discussion thread for adding the ot flex to PLR, hopefully using @koeng’s new server (Plr & ot)
i spoke with Daniel Abdoue today who is interested in writing this
1 Like
Ah, very nice. Does Daniel have any thoughts from the previous thread?
1 Like
asked him to join the forum so we should hear from the man himself soon!
Hi! Happy to be part of this community.
One particular aspect I’m hoping PLR will help me to do, especially with the proposed changes to the OT API, is create a global coordinate system such that an external multi-axis robot can also be used along with a liquid handler for sample handling. this way, I can tell the arm to pick up the sample that exists at deck slot A1 position 1 after the Flex has deposited solution onto the sample, and I can write wrappers such that a reference to A1 position 1 is defined as an x,y,z, and then the arm can handle the rest.
My use case is a bit different than usual; it’s materials research to optimize the best fabrication conditions for a semiconductor that is fabricated essentially by coating a piece of glass with a solution, drying it to evaporate solvents, and then putting the coated sample on a hotplate to anneal.
I think adding an “arm” module to PLR would be incredibly beneficial for expanding the complexity of automated labs that PLR can manage, such as mine. The eventual end goal idea being closed loop experiments where even characterizations are automated. I’ve been wanting to build an LLM based system very similar to what koeng has shown off recently, very interested to discuss more details and possibly collaborate on expanding the idea to general automated labs. I’ve been doing some work on automating paper analysis with LLMs to help narrow down potential candidates and conditions to try in an experimental setup, connecting information in one of the millions of ways to do RAGs or GraphDBs or whatever. (Unfortunately I’ll have to finish the software integration and scheduling for these robots before I can start thinking about that stuff though)
Since I’ve never used PLR before, it’d be great if I could get help accelerating the completion of hardware communication within PLR, and I’ll do my best to help provide support to you all by having access to the hardware
Here are some relevant resources regarding the Python API for the arm (mine is the xArm 6, though the software seems pretty easy to adapt to arms with less degress of freedom):
Please let me know if there’s anything you’d like me to try or provide further info on
2 Likes
Also, here’s the automated robot we currently use in our lab (though the Github is not completely up to date with all that’s changed, nor was the system made by me). User’s define their experiment in a jupyter notebook, and then that code schedules all the tasks required and generates the code. Then the user has to calibrate, setup all the hardware, and the robot performs the experiment. The same is also done for some basic characterizations, so theoretically end-to-end automation with experiment feedback is possible, and PLR would be a fastastic tool to make it all happen. Would love to make this new project I’m starting such that it’s easily replicatable and hardware agnostic
2 Likes