In the world of computer science, Robustness is a property of a computation network to deal with errors during execution/ run time and cope with erroneous input.
Formal techniques, such as fuzz testing and fault injection are essential to showing robustness (since this type of testing involves invalid or unexpected inputs). Various commercial products perform robustness testing of software analysis specifically those that deal with QA (Quality Assurance).
A system is considered to robust when it can continue operations even in the event of an error. As an example we can think of Bitcoin. Whenever a transaction takes place the BTC system must check that the balance of the sending address has enough assets to send. If the system detects that the sending address is trying to send 5BTC but only has 2BTC then an error will arise; but that error will not stop the other activity from happening on the network. Therefore, the Bitcoin network is considered “robust”.