You need to define what the results they're meant to achieve are clearly and completely unambiguously so they understand what they can control (essentially how they work, the order they develop things in and so on) and what they can't (usually what they're expected to deliver - both in terms of actual product and supporting materials such as progress reports - and when it's all meant to be delivered). You also need to let them know what resources they have - can they order high spec machines or order new software for instance or is that all decided?
I'd also ensure that one of their early deliverables was a schedule of completed milestones against which you could measure progress and agree with them what happens if they start missing milestones.
But I am slightly dubious about the idea that you're going to define version control, bug tracking and so on. Surely these are things you should let them decide? After all they're part of the process. Personally I'd state that they must have version control, centralised defect logging and so on but the mechanisms, tools and processes should be up to them.
It feels a little like you say you want to create a results only work environment but you don't quite trust them. If you're saying what you're going to do is create a ROWE then you need to make sure it's just that otherwise you're really only doing half the process and those situations rarely deliver the benefits people are hoping for.
After all, either you trust them or you don't but if you can't trust them to work out how to do version control which is frankly second nature to developers, you probably shouldn't be trusting them with the schedule which is normally a far less straightforward matter.