There really isn’t much difference in workflow, compared to the super-detailed robots.
You animate the parts. Where you see penetration, you animate the parts differently. Sometimes, you need to add extra twist or offset bones to make the pieces not intersect, and animate those.
If you need to support swappable parts, then you need to define envelopes for each of the parts, and what the keep-in/keep-out zones are for each of them, and then make all the animations based on those defined zones.
You could for example rig a simple character that just uses boxes or simple meshes for the keep-in zone for each defined part, and use that when animating, and then make parts of different detail level / style that still stay within the given zones, to build the final character.
There’s really no shortcut. Add enough bones, and make the animations such that they don’t interfere. And, because it’s robots, you typically want every vertex to only be weighted to a single bone.
With soft deforming characters, you weight the vertices to more bones, but then the problem is instead texture stretching, plus you often need to add separate twist bones to be able to animate different movements without too much distortion. (High-end character rigs will use deformers, too, for things like squashing biceps when folding elbows and such. But robots are rigid, and don’t need that, unless you have flexible tubing or somesuch.)