Our estimates get into trouble when trial means keep trying until it works. What decision should this pilot enable, and what amount of design effort are you prepared to spend finding out?
Put the unknown fixture work in a paid discovery stage. Small budget, actual deliverable. Buying the arm first won't make the nest design any less uncertain.
Anika, yes. Victor, include your time answering questions during the trial. Otherwise the operator looks independent while the programmer quietly becomes an unpaid second operator.
And agree what stops the discovery stage. A useful answer can be that your current tray supply makes the job unattractive. Nobody should need a successful demonstration to justify finishing the study.