The people I know that do cutting-edge research mostly do it because it's fun. That might be hard to imagine, but if you have great (government-funded) local hackerspaces where you can meet others, talk to them and built stuff together, it becomes a viable free-time activity. For example, I can schedule a Prusa (FDM) / Form (SLA) / Fuse (SLS) print job remotely and then quickly walk there to pick it up. If I need custom sheet metal or wood, I'll bring a USB stick with the DXF and then walk. That kind of infrastructure massively cuts down on iteration times. Plus it's great to get feedback in-person by other tinkerers when you pick things up.