There's a really cool international competition called the Space Settlement Design Competition, wherein teams of students collaborate to create really detailed designs for future human settlements in space. Think 70-page technical proposals for a luxury space cruise-liner to transport people between Earth & Mars in 2081, or for a research settlement in orbit around Venus in 2087. The competition has been running for 20+ years and receives 100+ entries each year. Here are 20 more proposals and I can help you get even more. There are also a lot of similar competitions.

So, there is a LOT of data here. This data isn't directly useful as actual plans (this is literally rocket science). But, the data does greatly capture what the students envisioned life in the distant future to be like. People are serious about winning the competition and so often put a lot of thought and research into what technologies they thought would become viable by then, and how life would be in the distant future.

It'll be really interesting if someone collated a lot of these proposals and then conducted a detailed analysis on how our imagination of technology in the distant future is evolving. You could answer questions like:

1. What did people in 2005 think about life in 2050? What did people in 2010 think? In 2020?
2. Were there any technologies that we anticipated would take a long while to emerge, but which emerged really quickly? Or perhaps things we expected to emerge quickly but are proving really difficult to find? What factors affect how well we predict the future?
3. What are forms of recreation could we think of? 
4. How did we think our relationships with robots would be like? What kinds of applications did we imagine?

Answering these questions could reveal biases and hidden mechanisms involved in our imagination of the future. It could lead to a very interesting scientific review, or possibly also an insightful newsletter like Pessimists Archive. As you explore, you might also come across some brilliant ideas buried in the files – ideas that one could bring it to life today. 

There's probably some existing research on these questions (using sci-fi stores as evidence, for example). This dataset would be different in that a lot more thought and technical consideration went into the content than it would've on other datasets.