What is expected when "novel" and "innovative" ideas are requested as part of a research grant application?

The purpose of adding words like "novel" and "innovative" is to weed out proposals that may be poorly researched and thus copying others or or involve repeating other works in a slightly different setting (or equivalent). The phrasings makes rejecting proposals easier since it provides a critieria (one among many) against which it can be judged. Funding agencies are also keen to see their funding go to research that can be judged to contribute "significant" results.

My experience says that what constitutes "novel" and "innovative" is difficult to assess in detail. In my field, environmental/earth science, some themes become "fashionable" or hot and signals these aspects, or rather absence means less likelihood to receive funding. These themes include finite element modelling (70s), acid rain(70s/80s) and climate change (currently). This can be seen as a communal will or interest to steer research in certain ways and so showing you can significantly contribute to these goals was/is more or less necessary to provide you with a good chance for obtaining funding.

So to define "novel" and "innovative" will be difficult. And, as a side point, your research either is or is not "novel" or "innovative", there are no degrees. You need to come up with ideas that are truly new (testing new grounds) or which promise results that are significantly advancing science but more importantly, you need to convince the reviewers in the funding agency about your case. This means to avoid "more of the same" proposals, to be sure your idea has not been worked on before (know your field).

A book, I strongly recommend for all is

Friedland, A., Fold, C.L., 2009. Writing Successful Science Proposals, Second Edition. Yale Univ. Press

The dumb (but maybe best) answer is: It's what the reviewers and the funding agency regard as "novel".

In some cases, this might be true for your example (e.g in medicine, it can be very valuable to investigate just an other aspect), but in most cases you should aim for something which is significantly beyond the current scope of the field (but still reachable). It should not be a straight-forward engineering approach (since this is development, not research).