The best practice in my opinion is to catch the Exception that occurs when it happens, then close the applications and either retry the section of the process or enter a closure sequence to end process.
I mean surround your workflow with a Try/Catch, so when something like that happens you can store the exception to a variable, then close applications, then either retry for a certain number of times or Throw the exception that was received with the Throw activity.
Take a look at the ReFramework model. While it’s missing some important components and is not very streamlined in my opinion, it can give you an idea on how you can take advantage of State Machines, Invokes, and Try/Catches to perform a robust process flow.
If you don’t like State Machines, you can perform the same model using a flowchart also.
If you can periodically check the Browser Response Status within your code, you could probably do something like this. you might need further validations if your robot is working in a high density environmen (multiple login) not to kill other user’s browsers.
Thank you sir, Ill also try that, I just need to be sure that if Excel stops working, I would kill the process and open it again. I guess that would work.