We do that exactly at my current place of employment. We call those projects the 100. Of those 100, 5 get a greenlight and of those 5, 1 launches. As a rule of thumb.
Generally, our customer-facing portion of NPD finds those 100 and usually asks one of our technical folks (designers) for assistance in developing prototypes for evaluation, which in turn generates some of the data when presenting to management to get a greenlight to the 5. That said, a designer can take the initiative to find a 100, I’m doing a couple myself at this point. And I am certainly capable to make prototypes (although I’m old and usually tell younger people to make me what I want) and I am certainly capable to taking them into the field for evaluation. But our customer-facing people have the contacts so I am going to lean on them.
My point is twofold. First is that we already have a culture of bringing “skunk works” to the light of day. I would hope most NPD organizations have something similar. Second, using the team is much more effective than going lone wolf. And as a manager I would question the efficiency of someone going lone wolf. Work smarter, not harder. Taking credit for something that could have been done better with the team is not a good thing in my book.
If your current culture makes doing NPD a challenge politically, you may want to rethink your organization.