Method for estimating a global uncertainty of a neural network

A non-transitory computer-readable medium on which is stored a computer program including instructions for estimating a global uncertainty of output data of a computer implemented main neural network of an at least partially automated robot, the computer program, when executed by a compute, causing the computer to perform the following steps:

• determining a first measure which quantifies to which extent current input data of the main neural network is following the same distribution as data which was used for training the main neural network;
• generating a second measure which quantifies the main neural network's certainty in its own prediction based on the current input data;
• ascertaining a third measure, based on an estimation of class-discriminative features in the input data and a comparison of these features with a class activation probability distribution;
• and
• determining the global uncertainty based on at least two measures out of the first measure, the second measure, and the third measure, wherein the at least partially automated robot is controlled based on the global uncertainty.