The first time you do things, you tend to do them wrong. If you're the first in the world then there are no examples to follow. Even if you're not the 1st you might be the first to face your particular set of circumstances.
1) Don't get rid of your experienced people
2) Don't let them block the efforts of new people to learn by doing.
3) Expect to fail, try to do it cheaply by testing things and investigating things long before it becomes critical for you to have a fully working, full scale version of them. The times when you have no critical requirement are golden opportunities to train.
4) Don't see failure as a reason to give up - it's the price of learning.
I was thinking that in a way, the economy, the manufacturing sector, project managers are rather like the 4th armed service. They need to be in trained, go on exercises, be prepared continuously to output all the things that might be needed.
e.g. take the Bloodhound SSC project. That seems like great training for everyone involved in the development of a new product and by military standards it is probably quite cheap. There need to be lots of projects to develop things if you are to build up the skills of people in doing it.