R Regex / gsub : How to collapse spaces in a string

I have a vector of sentences that were scanned from handwritten documents. In the process there were some spacing problems like this:

 The d og is br own.

I was curious if there was a way to generically take any pattern with '_x_' or space-character-space and collapse the second space like this:

The d og is br own.  --> The dog is br own.

I'm only worried about a single character between the spaces ('_x_' NOT '_xx_').

Any suggestions?