I would. Give the experiment its own labelled materials and keep its records separate. A nice robot result shouldn't accidentally become permission to handle real samples.
0分
I'm exploring a sample-handling experiment and the SmolVLA article is on my reading list. I'd start with empty capped tubes and a fixed rack, without actual samples. My worry is that a successful transfer says very little about contamination or traceability. Would you keep the learning experiment completely separate from the lab workflow until those questions are answered?