a processor configured to:

detect an environment in which a first automated machine and a second automated machine are located, wherein the first automated machine is affiliated with the second automated machine in a group;

determine an action taken in the environment by the second automated machine;

determine an action expected of the second automated machine;

compare the taken action with the expected action;

determine an accuracy score of the second automated machine based on the comparison, wherein the accuracy score comprises a static prediction of the expected action and a runtime result of the taken action, wherein the static prediction is based on a trained task performance model that predicts the expected action;

remove the second automated machine from affiliation in the group based on the accuracy score of the second automated machine and based on a deviation between the static prediction and the runtime result, wherein the deviation comprises an extent to which the static prediction deviates from the runtime result;

update the trained task performance model to an updated task performance model based on the deviation; and

control movements of the first automated machine or the second automated machine based on the updated task performance model.