Speak to them privately, away from any outside influences. This seems like a pretty reasonable way to ensure someone isn't coerced. I suppose there are complex situations (e.g. someone is holding your children hostage) where this may not be sufficient, but it's unclear why someone would go to that trouble to coerce someone to lend themselves to such an experiment. This seems like mostly a non-issue.
> how do we validate that volunteers truly comprehend the risks?
Have a psychiatrist/psychologist interview them. Give them a long cooling off period, so that it's not a spur of the moment decision. Require the sign-off of the psychiatrist/psychologist, or panel thereof before approval.
> how do we ensure that volunteers are of sound mind and not suicidal?
Why do you need to do that? If someone wants to kill themselves, we might as well let them contribute to humanity in so doing.
> if someone is gravely injured or disabled in the experiment, is the institution prepared to care for them for the rest of their life?
We could decide that they must be, if we felt that was necessary.
> if we offer compensation, how do we account for the fact that people of lower socio-economic status will be more likely to take that compensation for the same risk?
Why is that a problem? People often point this out, but I don't really get it. Let's phrase it another way: Why shouldn't we give poor people agency over themselves and their situation? The fact that poorer people are more get the option to do this thing that could improve their situation seems like only a great thing to me. This only gives them a new choice that they didn't previously have - it's hard for me to see how that could be a bad thing.