For now, sure, obviously it'll be more natural to use the mediums that are pervasive on commercially available devices that are backed by several industries-worth of engineers and developers, compared to some systems hacked together by handful of grad students in some lab somewhere or hardware startups in a very nascent stage. Hardware is hard, and these platforms are just beginning to explore the possible application spaces of which there is still a lot of research to be done. My personal view on many of these telepresence platforms is that they're basically roving tablets with limited embodied expressiveness and no manipulation capability, which are whole subfields of research in-and-of themselves, and that addressing these limitations will eke out more value of telepresence robots in the future. The point of research (ideally) shouldn't be an immediate VC-backable commercial application, but rather to explore possible futures viable 5-10+ years down the road.

Thus robots are not meant as a replacement for virtual learning, but as a way for students to follow the normal education when they are temporarily away.

For the last two years, I have been taking an MBA, and my school had thus robots. When students could not come to class, they could still attend by using a AV1. It is not so easy for the teachers to also have to do a virtual meeting simultaneously as teaching the regular class, so the alternative is often that the teachers just send out their slide deck to those that were not present, so the student can self-study. The robot also has batteries, so we took it with us to group work, lunches, and parties. Here is a picture of my group during the second covid surge, hammering away on our digital business model assignment: https://www.instagram.com/p/CCGS2lxhgcU/ .

Maybe serve as a physical reminder for your friends that you are still part of the class to make it easier when you then get back?

I think I agree with this being over-hyped, but trying to look for reasons it might be a good idea.