I think that's closer to Cheney's algorithm (which is one of the easier ones to make bounded-latency actually).

I'm new to these concepts, so please forgive me for coming up with something that already exists.

I guess then the question becomes why would the roadmap avoid such an implementation?

C4 is a well known pauseless GC, the paper is good and digging through the citations will get you pretty caught up on the theory and practice:

http://www.azulsystems.com/sites/default/files/images/c4_pap...