Just kind of spitballing

For starters, you design a robot whose design is to take, for example, 100 steps forward, then take 100 steps back. Then you could:

Have it take a scoop of soil before it turns back, you now have a sample to test

Put some kind of chemical test strips (litmus paper, water quality, etc.) on it and send it towards a puddle of water you want to test. It splashed into the water and comes back, now you can see what the results on those test strips say

Not all electronics are so sensitive to radiation, and to some extent they can be shielded. Building a whole digital robot that’s hardened against radiation would be difficult and expensive. Sticking a couple radiation hardened sensors on an otherwise dumb pneumatic robot that doesn’t need to be hardened would be much cheaper.

Send the robot in with analog measuring tools- thermometers, barometers, film cameras (radiation can expose film, so picture quality may not be great, but it’s better than nothing, and just seeing how exposed the film is could be used to get a rough idea of the radiation level) etc.

Say there’s a big rock, and you need to know what’s behind it. Nowhere in the safety zone has a good viewing angle, and it’s either unsafe to fly over the area or there’s too much tree cover so you can do aerial photography. So you stick a mirror on the robot and send it out somewhere behind the rock off to the side o bit. Now you can look at the mirror through some binoculars or a telephoto lens and see what’s behind the rock in the reflection.

These are just a couple off the top of my head ideas as a layperson, I’m sure that a scientist or engineer actually doing this kind of work whose entire job is to think about this could come up with plenty of other good ways to use this sort of thing.

Electronics definitely make things easier, but we’ve had people doing science for millennia before we figured out how to do anything particularly useful with electricity.