This makes a lot of sense -- Word Lens was always a very impressive bit of real-time image manipulation, but the "translation" left an awful lot to be desired. Google are in the best position to improve that. The only downside is that integrating it with Google Translate will require an "always-on" connection, so it may be less travel friendly in the long run.
In the article it is stated that the app can work even without access to the internet. In that respect it is one step ahead of Google translate, but otherwise it would be logical for Google to attempt to connect these two pieces of software (and adding a voice recognition module to the complex - hopefully a better one than the dubious attempt at Youtube)
It does indeed, but it's rubbish. It's been improved a bit since the first public release, but it's still really bad at determining sense from context.