Since Pega Robotics is not a screen scraper, it must be installed on the machine with the applications being automated. To achieve this use case, you could write one automation to launch RDP and login to the session. On the remote machine, you can have Runtime launch on startup to load a second automation to get the data from your web page and save it to a central location. Your first automation would watch the remote location and continue when the data has been saved there.
The proper way though, would be to use the RPA Service and Robot Manager as it can login to a remote machine. You would handle everything through the case in the Pega platform while your bot would simply return the data from the page to the case for the remaining processing.