For a small indie developer, with limited funds/time (and maybe more of a focus on 'making something cool' than 'making something profitable'), trying to go cross-platform from the start could be counterproductive. It takes a lot of effort to engineer solid cross platform tools and tech (different graphics APIs, endianness, input devices, and more) - time that could be spent on the more creative side of game development.

But you probably want to make sure you've got a great game that works really well on one platform before worrying too much about getting it onto as many platforms as possible! If the game is a flop, there's no point wasting time and effort making it a multi-platform flop, is there?

If you're coding in C/C++, mostly from scratch, then as long as you keep the code fairly modular and make sensible decisions about data formats and middleware/libraries, then supporting other platforms later shouldn't be too painful.

If third-party cross-platform tech/tools (e.g. Unity) is an option for your project, then it's certainly worth considering.

The main 'problem platforms' for indies would seem to be Xbox360 Indie Games (C# only, limited network access, etc), and possibly Android (massive differences in device performance/screen size/input devices). If you're determined to support these, expect a more sizable porting job, or plan to focus on them exclusively.